WORLD METEOROLOGICAL ORGANIZATION

TROPICAL METEOROLOGY RESEARCH PROGRAMME (TMRP) COMMISSION FOR ATMOSPHERIC SCIENCES (CAS)

TROPICAL WORKSHOP TOPIC REPORTS METEOROLOGY SIXTH WMO INTERNATIONAL WORKSHOP ON RESEARCH TROPICAL CYCLONES (IWTC-VI) PROGRAMME REPORT SERIES (SAN JOSE, Costa Rica, 21-30 November 2006) TMRP No. 72

World Meteorological Organization – Geneva - Switzerland

WORLD METEOROLOGICAL ORGANIZATION

TROPICAL METEOROLOGY RESEARCH PROGRAMME (TMRP) COMMISSION FOR ATMOSPHERIC SCIENCES (CAS)

TMRP 72

WORKSHOP TOPIC REPORTS SIXTH WMO INTERNATIONAL WORKSHOP ON TROPICAL

CYCLONES (IWTC-VI)

(SAN JOSE, COSTA RICA, 21-30 November 2006)

WMO /TD No. 1353

Foreword

This volume contains reports from Topic Chairs and Rapporteurs for the Sixth International Workshop on Tropical Cyclones (IWTC-VI). As in IWTC-V, the International Committee has chosen as a Special Theme for this Workshop, which is on the “Quantitative Forecasts of Tropical Cyclone Landfall in relation to an Effective Warning system”. This topic is of significant interest to all tropical cyclone forecasters and other disaster management and mitigation agencies. Various aspects of track forecasts as well as observations and forecasts of wind, rainfall, storm surges and other hydrology-related issues will be addressed. The other main Topics of the Workshop include: (1) Tropical cyclone structure and structure change, (2) Tropical cyclone formation and extratropical transition, (3) Tropical cyclone motion, (4) Climate variability and seasonal prediction of tropical cyclone activity/intensity, and (5) Disaster mitigation, warning systems and societal impacts. The main purposes of the Topic reports are to summarize progress since IWTC-V and to make recommendations for the IWTC-VI participants to consider. The preparation of these reports is a daunting task and requires a significant amount of work by Topic Chairs and Rapporteurs, in collecting and collating input from their team members and in the actual writing of the reports. For all their efforts as well as the contributions from their team members, we would like to express our sincere appreciation. The substantial increase in our knowledge in various areas of tropical cyclones has led to a concomitant increase in the “thickness” of this volume, which is over 500 pages, not to mention supplementary material (colour figures) included in the CD that accompanies this print volume. With this wealth of information, we are sure that this volume will serve to bring about a fruitful outcome of IWTC-VI. Johnny C L Chan Chiu Ying Lam Co-Chair, IWTC-VI

TABLE OF CONTENTS

Topic Title Topic Chair / Rapporteur Page

Topic 0 Special Theme: Quantitative Forecasts of Tropical Cyclone Landfall in relation to an Effective Warning system

Jim Davidson (Australia)

Report of Topic Chair Jim Davidson (Australia)

1

0.1 Track forecasts Lixion Avila (USA)

12

0.2 Observations and forecasts of wind distribution Sai-Tick Chan (Hong Kong, China)

19

0.3 Observations and forecasts of rainfall distribution Lianshou Chen (China)

36

0.4 Observations and forecasts of storm surges Shishir Dube (India)

43

0.5 Observations and forecasts of hydrology-related issues

Kang Thean Shong (Malaysia)

56

Topic 1 Tropical cyclone structure and structure change

Hugh Willoughby (USA)

Report of Topic Chair Hugh Willoughby (USA)

62

1.1 Environmental effects on tropical cyclone structure and Structure Change

Liz Ritchie (USA)

68

1.2 Tropical cyclone Inner-core dynamics Jeffrey D. Kepert (Australia)

79

1.3 Air-sea interface and oceanic influences Nick Shay (USA)

120

1.4 Operational techniques in defining tropical cyclone structure

Mark Lander (USA)

151

1.5 Operational guidance and skill in forecasting structure change

John Knaff (USA)

160

Special Focus Topics 1a Tutorial on the use of satellite data to define tropical

cyclone structure Chris Velden (USA)

185

1b Field experiments related to tropical cyclone structure (CBLAST & RAINEX)

Peter Black & Shuyi Chen (USA)

214

Topic 2 Tropical cyclone formation and extratropical transition

Pat Harr (USA)

Report of Topic Chair Pat Harr (USA)

256

2.1 External influences on formation Bill Frank (USA)

261

2.2 Internal influences on tropical cyclone formation Mike Montgomer (USA)

266

2.3 Operational forecasting of tropical cyclone formation Jeff Callaghan (Australia)

288

2.4 Observing and forecasting of extratropical transition Jenni Evans (USA)

305

2.5 Physical processes and downstream impacts of extratropical transition

John Gyakum (Canada)

320

Special Focus Topics 2a The Catarina phenomenon Pedro Silva Dias

(Brazil) 329

2b Special Focus Session THORPEX: a focus on tropical cyclone related research

Jim Abraham (Canada)

361

Topic 3 Tropical cyclone motion Russ Elsberry

(USA)

Report of Topic Chair Russ Elsberry (USA)

373

3.1 Advances and requirements for operational track prediction

Tsz-Cheung Lee (Hong Kong, China)

376

3.2 Improvements in understanding and prediction of tropical cyclone motion

Sim Aberson (USA)

390

3.3 Targeted observation and data assimilation in track prediction

Chun-Chieh Wu (PSA)

409

Special Focus Topic 3a Sharing experiences in operational consensus

forecasting Andrew Burton (Australia)

424

Topic 4 Climate variability and seasonal prediction of tropical cyclone activity/intensity

Chris Landsea (USA)

Report of Topic Chair Chris Landsea (USA)

442

4.0 The recent active Atlantic hurricane seasons Max Mayfield (USA)

446

4.1 Variability of tropical cyclone activity/intensity on intraseasonal and interannual scales

Chang-Hoi Ho (Republic of Korea)

456

4.2 Possible relationships between climate change and tropical cyclone activity

Tom Knutson (USA)

464

4.3 Short-term climate (seasonal and intraseasonal) predictions of tropical cyclone activity/intensity

Suzana Camargo (USA)

493

Special Focus Topic

4a Updated statement on the possible effects of climate change on tropical cyclone activity/intensity

John McBride (Australia)

500

Topic 5 Disaster mitigation, warning systems and societal impact

M C Wong (Hong Kong, China)

Report of Topic Chair M C Wong (Hong Kong, China)

505

5.1 Evaluating the effectiveness of warning systems Woo-Jin Lee (Republic of Korea)

510

5.2 Factors contributing to human and economic losses Roger Pielke Jr. (USA)

533

5.3 Mitigation strategies and community capacity building for disaster reduction

Linda Anderson-Berry (Australia)

556

Special Focus Topic

5a Report to the IWTC-VI on the PROGRAM FOR IMPROVEMENTS TO HURRICANE INTENSITY FORECASTS AND IMPACTS PROJECTIONS (HiFi)

Greg Holland & Roger Lukas (USA)

564

1

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES Special Theme Topic 0: Quantitative Forecasts of Tropical Cyclone Landfall in relation to an Effective Warning System Topic Chair: Jim Davidson, (Australian) Bureau of Meteorology 0.1 Track Forecasts 0.2 Observations and Forecasts of Wind Distribution 0.3 Observations and Forecasts of Rainfall Distribution 0.4 Observations and Forecasts of Storm Surges 0.5 Observations and Forecasts of Hydrology-related Issues Abstract This Workshop report summarises developments in the 4 years since IWTC-V in providing quantitative forecasts of tropical cyclone landfall in relation to an effective warning system. Here the landfall phase is defined as that period of a cyclone’s lifecycle bounded by the immediate approach of the cyclone to the coast through to the early part of its transition over land. The five rapporteurs have reported on monitoring, scientific, and modelling advances since 2002 and all have highlighted the established fact that landfall parameter prediction is highly dependent on the cyclone’s track. At a particular location, even small deviations in track can result in vastly different impacts from wind, storm tide, precipitation and flooding. Most deaths in cyclones continue to be caused by flooding and landslides. With the steady growth in coastal population and infrastructure, we are witnessing what amounts to a quantum leap in vulnerability (and therefore risk) in many cyclone prone areas, irrespective of trends in cyclone numbers and intensities. Evacuation times have increased accordingly. More focus is directed in recent years to gaining a better understanding of the smaller scale wind features in land-falling cyclones as now regularly evidenced by prominent streaks in the surveyed damage pattern. Also attracting global attention are the apparent anomalies in the operational application of the Dvorak technique for estimating cyclone intensity. In accord with an IWTC-V recommendation, the wind conversion factors appearing in the Global Guide on Tropical Cyclone Forecasting are being updated. Any one or possibly all of these developments could ultimately have a bearing on the forecasting of a cyclone’s landfall parameters and category scales. Prospects for further improvements in quantitative forecasts of cyclone landfall appear to rest on greater availability of consensus/ensemble forecasts of not only track and intensity but also precipitation. A multi-global model approach is preferred. Significant forecasting gains can also be realised through the continuing development and distribution of parametric models for a range of landfall parameters. As resources allow, these initiatives should be underpinned by field experiments, demonstration projects and various capacity building activities. Remote sensing data, especially from satellites, will remain critical to the forecast process.

2

(A) Effective Warning System To set the scene, what does constitute an “Effective Warning System”? Extract from WMO PWS-13 “Guidelines on Integrating Severe Weather Warnings into Disaster Risk Management” – Jim Davidson and M. C. Wong – 2005: “The primary objective of a warning system is to empower individuals and communities to respond appropriately to a threat in order to reduce the risk of death, injury, property loss and damage. Warnings need to get the message across and stimulate those at risk to take action. Effective inclusion of the severe weather warning system in a risk management plan relies on NMHSs fully appreciating the needs of a multi-cultural, economically stratified and often mobile community – and the community understanding the hazard, its vulnerability and the most suited protective action to take. Greater focus towards disaster mitigation also means:

• Further increasing the emphasis on extending the lead time of warnings, • Improving the accuracy of warnings at varying lead times, • Presenting warnings in different formats (including graphical products), • Satisfying greater demand for probabilistic forecasts, • Better communication and dissemination of warnings, • Using new technologies to alert the public, and • Better targeting of the warning services to relevant and specific users (that is, the right information

to the right people at the right time at the right place). Extract from Final WMO Report on Meeting of Expert Team on Public Weather Services in Support of Disaster Prevention and Mitigation – June 2006: “The scope of disaster prevention and mitigation extends beyond the Planning phase to also include relevant NMHS actions and responsibilities during the Preparedness and Response phases of Disaster Management. In partnership with emergency services authorities, prevention and mitigation by NMHSs during the 3 phases are broadly defined in Figure 1 as being: • Undertaking hazard assessments

utilising severe weather databases during the Planning phase;

• Conducting public awareness programs explaining the hazard risk during the Preparedness phase; and

• Operationally including community protective action advice in warnings during the Response phase.”

3

(B) Developments since IWTC-V in 2002 Extract from IWTC-V Final Report - Tropical Cyclone Landfall Processes “The major loss of life in recent decades has been from inland rainfall and its’ associated flooding, land slips and mud slides.” The GOOD report card in 2002: “The available observing systems and techniques have reached the level of sophistication to enable detailed analysis and research into the related processes and effects on the cyclone structure and associated impacts. We have seen an enormous increase in relevant data from the USA field programs and increasing operational use of remote sensing systems, both satellite and land-based, which provide a wealth of information for research activities. Numerical models with higher resolution allow direct calculation of mesoscale (and even microscale) processes, and lead to improved parameterizations, and thus provide a capacity for both improved research and future operational forecasts. Innovative basic research approaches are improving our understanding and capacity to analyze important features of the boundary layer wind field. In addition, the impacts/response and meteorological communities have moved closer together in improving the warning message and its communication.” The NOT SO GOOD report card in 2002: “We have very little real skill in forecasting the intensity and structure of tropical cyclones as they approach land, especially when rapid intensity changes occur (up or down), and no skill in predicting any details of the rapid structural changes that occur at landfall. No real skill exists in forecasting tropical cyclone rainfall, and in many cases such forecasts are little more than a general indication of rainfall conditions. This deficiency is the major limitation to hydrological modelling of the run-off and flooding that ensues. Whereas the storm surge modelling is relatively sophisticated, the high dependence on the tropical cyclone track and the wind structure and the poorly modelled interactions with river flooding, are critical limitations.” The criticality of TRACK forecasts in 2002 (and TODAY): “Overlaying all of these aspects is the great dependence on track forecasts, where even relatively small track errors can result in substantially different rainfall, storm surge, precipitation, and localized wind-field changes.” Summarising in 2006: In a very general sense, not a great deal has changed since IWTC-V in regards to marked improvement in the “quantitative forecasts of tropical cyclone landfall”. There is little doubt however that the much greater range/quality of information today, especially from weather satellites, has enabled forecasters to better monitor the cyclone track, intensity and structure. The following forecasting strengths and weaknesses still remain, as well demonstrated in the Rapporteur Reports. Relative Strengths in forecasting landfalling tropical cyclones • Track forecasting (but note however that small deviations do matter) • Hydrological and Storm Surge modelling (but still dependent on inputs) Relative Weaknesses in forecasting landfalling tropical cyclones • Forecasting cyclone intensity, structure, and structural change • Forecasting spatial and temporal rainfall patterns • Modelling wave action (including wave set-up and wave run-up) • Modelling the combination of riverine flooding, storm tide and waves

4

Of relevance to this Session are the major outcomes of an International Workshop on Tropical Cyclone Landfall Processes held in Macau, China in March 2005. To quote directly from the Workshop Summary Report: “The participants gave the following ranking of priorities in regards to improving tropical cyclone landfall forecasts: (1) Further improvements in track landfall forecasts; (2) Improved predictions of tropical cyclone-related precipitation following landfall; (3) Advances in understanding and predictions of structure and intensity during and following landfall,

and from tropical storm stages to extra-tropical transition; and (4) Further applications of storm surge models, including improved specification of the

meteorological forcing.” Also very worth noting are the key findings of the Workshop Working Group for Tropical Cyclone Landfall Impacts. A series of questions and the associated unedited responses are summarised below. (1) What have been the chief advances in understanding and forecasting the various impacts of a TC landfall? • Track forecasting • Numerical modelling • Remote sensing data especially satellite • Availability of radar data on the Internet and TV weather channels has improved chances of people

responding appropriately • Greater knowledge of TC structure eg concentric eye walls, rain bands mainly through better

observations • More graphical products have benefited the community • More risk/vulnerability studies • Better communication between NMHSs, media and disaster managers • Inclusion of “protective action advice” in warnings • Better understanding of wave action and its components (2) What are the prioritised requirements related to the impacts of TC landfall? • Better topographic and bathymetric data for modelling • Better modelling of the combined effects of storm tide, wave action and riverine flooding • Better modelling/monitoring/validation of inland flooding from storm tide (as water kills) • Better modelling of wave action especially for vulnerable islands • Better historical records for both mitigation and warning • Better surface observational platforms • Better information on TC impacts through developing and maintaining databases • More risk assessments of low probability high impact TC events • Greater use of TC forecast graphics (3) What are the meteorological requirements for improved storm surge, hydrology and wind decay forecasting? • Good return period (or exceedance probability) TC hazard maps • More surface observations (land and ocean) available in real-time • Better temporal and spatial rainfall forecasts • Better description of the TC wind profile • Greater resource focus on the dangerous/destructive elements • Wind decay rate is important and is often influenced by neighbouring weather systems • Greater use of GIS in mapping the TC hazards

5

• Influence of tidal cycle on storm tide prediction so better timing of landfall is critical – keeping in mind that there’s both a directional and timing element to a TC track forecast

• TC speed after landfall is also critical as this could well have a bearing on both rainfall amounts and wind damage eg a slow moving TC will often produce relatively high rainfalls and greater wind damage

(4) What are the gaps in observations, models etc that must be filled to make progress? • More aircraft observations eg dropsondes, doppler radar • Need to ingest radar data into numerical models • Greater sharing of data/techniques between nations especially in the same region • The question was then asked – how many global numerical models does the international

meteorological community really need? • More radars – both doppler and conventional • More use of high resolution coupled (in every way) numerical models • Greater availability of ensemble prediction systems, consensus forecasts, TC strike forecasts and

probability forecasts • Development of a suite of parametric models for all or most TC hazards • More attention to developing techniques for estimating the interaction between storm tide, wave

action and riverine flooding (5) What are the opportunities for working with societal impacts experts to improve warnings and responses to warnings of TC landfall? • Establish an inventory of “where we are now” with the involvement of social scientists • Many social scientists out there working in the field of natural disasters but very few aligned with

NMHSs • Employ social scientists (in collaboration with emergency managers and maybe the media) to

construct the “protective action advice” statements in warnings • Social scientists would arguably improve the communication, perception, understanding, and

response to warnings • Economists could work out the costs/benefits of disaster mitigation measures, estimate avoidable

losses etc • More international collaboration is needed between “TC hazard” social scientists • Social scientists could assist in mitigating the overconfidence which is sometimes exhibited by

emergency managers • Important to continue to include “TC Impact” session(s) at IWTCs (6) If there is to be an international or multi-nation program, what aspects of TC landfall impacts should be the central foci? • Simulator model – a significant TC event in one part of the world could be transposed to another

region or country to test those components of the total warning system from warning to response • This has the advantage of being more realistic than the standard table-top exercise with a synthetic

TC • Actual evacuations could be factored into the exercise • The weakest link in the total warning system is more likely to be in the response area than with the

warning • The simulator model aside, field and community surveys should be included in any TC landfall

impact program/exercise • The SW Pacific and the Caribbean were singled out as the parts of the world (where TCs are

experienced) that could benefit most from programs/exercises

6

(C) Rapporteur Report Summaries 0.1 Track Forecasts (Lixion Avila) The rapporteur noted the steady improvement that has been realised in tropical cyclone track forecasts over the past 10 to 20 years. The reduction in track error could be attributed to various factors, including the improved quantity and quality of observations from a suite of observational platforms ranging from satellites to dropsondes, the use of improved data assimilation techniques (particularly for unconventional satellite observations), and significant advances in the resolution and physics of NWP models. This improvement has been generally reflected in both a reduction in the average length of coastline placed under cyclone warning and an increase in warning lead-time.

Centres issuing tropical cyclone warnings must carefully balance the need to minimize the negative impacts of over-warning against the need to safeguard lives and property. In many cyclone prone areas, the continual increase in coastal population and infrastructure without significant expansion of evacuation routes has resulted in the need for increased evacuation lead-times. This largely explains why greater attention is being placed on translating forecast improvements into increasing the lead-time of warnings as opposed to reducing the warning area. A number of centres are now providing a pictorial “cone of uncertainty” to display the uncertainties in a track forecast, using either historical forecast track errors or single or multi-model consensus/ensemble output. RSMC Miami has also begun to use “wind speed probabilities” graphics to even better convey the uncertainty in track, intensity and size forecasts. 0.2 Observations and Forecasts of Wind Distribution (Sai-Tick Chan) The rapporteur noted the inability of NWP models to be run at a resolution high enough to adequately resolve the cyclone structure, especially in an operational mode. Only a limited number of objective guidance techniques have been specifically designed for the task and statistical methodologies continue to be the mainstay in defining/forecasting the wind distribution. Research is continuing into the structural and wind distribution changes arising from land-sea contrasts, in particular the influence of land-induced asymmetric friction on the boundary layer winds, and the effects of the moist processes in introducing asymmetries in structure at landfall. In recent years due to an increased number of phenomena reports, more research is focussing on the damaging fine-scale surface wind features such as boundary-layer rolls, small-scale spiral bands and vortices, wind streaks, and terrain-induced wind accelerations (both up-slope and down-slope). The GPS dropsonde has lead to a greater understanding of cyclone structure, including eyewall replacement cycles in the more intense systems. Extreme gusts have been reported in both horizontal and vertical wind components. Such events contribute greatly to the extensive wind damage that can result from a landfalling cyclone that cannot be adequately represented by the Saffir Simpson Scale and similar broad-based cyclone category systems. The justification for using basin-specific wind-pressure relationships was challenged, and it was argued that the variability in wind-pressure balance within and between individual cyclones is likely to be much greater than any regional “average”. Therefore it is considered important to incorporate storm spatial scale and profile shape to allow for the expected natural variability in the wind-pressure balance. The Holland B parameter is a widely applied indicator of the wind-pressure relationship. One of the more recent studies provides a basis for utilising the readily available operational parameters of size, latitude and environmental pressure to enable forecasters to adopt specific wind-pressure pairings on a case-by-case basis. Notwithstanding the increasing availability of remotely sensed data (e.g. scatterometer) for estimating

7

surface wind speeds, there remains a critical need to maintain and expand surface wind measurement networks in all areas prone to tropical cyclone impacts. Verification of forecast/modelled winds must remain a priority if real improvements in techniques and procedures are to be realized. Standardisation of wind sampling is also essential.

“Wind Speed Averaging Guidelines” are currently being reviewed in a study sponsored by WMO-TCP, with the goal of updating the relevant mean wind-gust factor table in the Global Guide for Tropical Cyclone Forecasting. It is expected that the results of that study will be presented at IWTC-VI. Parametric wind field modelling is not generally available to cyclone forecasters. In the absence of aerial reconnaissance, Dvorak is often seen as the only practical operational technique for estimating intensity. Parametric modelling does rely on having at least one or two nearby surface wind and/or pressure observations but can be readily applied within a simplified modelling context to add significant value to satellite-only intensity estimates. The practical utility of parametric models has been boosted in recent times by the emergence of scatterometer data. SHIPS (Statistical Hurricane Intensity Prediction Scheme) is a relatively successful statistical-dynamical model used in some centres for intensity forecasting, and several recent modifications have improved the technique. The rapporteur made a number of recommendations that included (1) more research into structural changes at landfall and extreme localised wind gusts, (2) greater operational use of parametric wind field models, (3) improved verification of forecast/modelled winds, (4) expanded network of wind/pressure observations in cyclone prone areas, and (5) standardisation of wind measurement and reporting. 0.3 Observations and Forecasts of Rainfall Distribution (Lianshou Chen) The rapporteur noted that remotely sensed observations are important in estimating tropical cyclone rainfall QPE. Included here are TRMM, SSM/I, AMSU and IR/VIS satellite data as well as (conventional and Doppler) radar reflectivities. A dense network of rain gauges is also very desirable for rainfall calibration of the remote sensors. Not to be overlooked is that the basic ingredient of tropical cyclone rainfall is moisture supply, best estimated from radiosondes, satellite images and/or radar reflectivities. A range of QPF techniques have been employed with landfalling tropical cyclones. These include limited area or mesoscale models, with remotely sensed data assimilation improving initialization of the model and the rainfall forecast accuracy. Single and/or multi-model ensemble rainfall forecasts are available in a small number of centres, and showing good potential. Statistical, combined statistical-dynamical and empirical approaches are still commonly used. A significant advance in recent times is the Very Short Range Forecast (VSRF) that provides 6-hour quantitative precipitation forecasts updated every 30 minutes. The VSRF is based on extrapolation of the latest observed precipitation for the first 3 hours and then model results are combined with extrapolation for the second half of the period. The rapporteur made a number of recommendations that included (1) both a field experiment and a demonstration project on landfalling cyclones, (2) a global NWP rainfall comparison project, (3) roving seminars, special workshops and symposia on rainfall forecasting, and (4) greater operational availability of satellite data used in forecasting rainfall. A greater research effort also needs to targetted at calibrating satellite data.

8

0.4 Observations and Forecasts of Storm Surges (Shishir Dube) The rapporteur noted that the prediction of storm tide and the extent of coastal inundation depend critically on the prediction of track, intensity and the spatial structure (wind, pressure) of tropical cyclones. There has been little new published material on the subject of the modelling and forecasting of storm tide since IWTC-V. This may be due to a number of factors including (1) the hydrodynamics of storm tide generation and propagation is relatively well established, (2) the accuracy of storm tide predictions is largely limited by the meteorological inputs, and (3) the implementation of new regional models is limited by scarcity of resources. The advent of powerful PC-based workstations has established a trend to run storm surge models operationally in real-time. This has been recently augmented by the adoption of rapid assessment parametric storm tide forecast models and probabilistic storm tide forecast models in at least a couple of centres. While the probabilistic forecasting of cyclone track has become more widespread through the combination of ensemble NWP tracks, the extension to wind probabilities and especially storm tide forecasting is yet to be widely adopted. The advantage of probabilistic storm tide forecasting is that the full range of parameter possibilities can be explored (track, velocity, intensity, scale, tide, timing etc) and the forecaster can focus on those specific meteorological parameters that will have the greatest effect on the storm tide forecast in a defined coastal zone. Typically, numerical storm tide modelling systems incorporate simplified (analytical, parametric) descriptions of the wind and pressure fields of tropical cyclones. While such representations are consistent with the level of detail currently provided by the Dvorak technique for example, more sophisticated models have become available in recent years that have the capacity to improve the representation of wind and pressure fields. Considering the steady advancement of NWP models, the feasibility of applying mesoscale NWP model results to storm tide modelling should be pursued, as an improvement in storm tide forecasts is likely. Some “bogussing” of the initial conditions in the NWP model would be desirable. While there may be advantages in developing or using more complex coupled ocean and atmospheric models for improving cyclone intensity forecasting, they can be costly and the advantages to storm tide forecasting are likely to be less critical, especially if coastal bathymetric data is of poor quality. Simplified analytical wind and pressure models continue to perform satisfactorily provided uncertainty in parameter estimation is considered. Wave set-up is technically a component of storm tide and therefore access to a suitable wave model is also required or a parametric approach adopted. The rapporteur made a number of recommendations including (1) ensemble and probabilistic methods considered for operational use, (2) high resolution mesoscale NWP models to provide meteorological input investigated, (3) robust and reliable operational storm tide techniques developed, (4) further studies undertaken in estimating wave set-up, (5) coupled ocean-river models developed where applicable, (6) archiving of actual storm tide observations/reports for model calibration, (7) improved response through storm tide education of disaster managers, and (8) capacity building through workshops/seminars.

0.5 Observations and Forecasts of Hydrology-related Issues (Kang Shong) The rapporteur noted that the severity and extent of hydrological related tropical cyclone disasters, particularly flooding, landslides and debris flow appears to have increased in recent years. Thus an early warning system for hydrological related hazards is becoming more critical. The main challenge here is forecasting the spatial and temporal rainfall pattern. Some of the essential components of early warning systems for hydrological related disasters are flood, landslide and debris flow hazard mappings, flash flood and sediment disaster forecasting and warning, and flood forecasting model evaluation. Hydraulic models are usually not run operationally, but

9

nevertheless the hydraulic model results are used in refining the flood hydrological models. Forecasting of floods produced by land falling cyclones still has a high degree of uncertainty due to a number of factors. Often the model is adequate but the hydrologic and/or hydraulic data and the precipitation network are either non-existent or poor so calibration of the model is difficult. Determining the flood producing potential (QPF) of an approaching cyclone is a related limitation. There is a need to also continue in improving hydraulic and hydrological flood modelling, and this necessarily includes consideration of a tighter coupling with tidal and meteorological modelling outputs. Both ensemble rainfall forecasts and (very) short term QPF are desired by forecasters and hydrologists alike. Allowance should be made for gridded rainfall inputs to hydrological models to take advantage of improved spatial and temporal rainfall analyses from NWP models. Hydrological model domains could then be digitised to enable gridded rainfall inputs. Intense rainstorms produced by tropical cyclones often cause severe landslides and debris flows, which have claimed many lives and properties. It is imperative that landslide and debris flow high resolution forecast models are developed, and for that purpose dense rainfall networks and strategically placed radars are essential.

The rapporteur concluded that although some of the existing hydrological models are simplified conceptual representations of rainfall-flood response and fail to model the complexities of the land-based water cycle, the simplified models do provide reasonably good operational forecasts. Experience in flood forecasting was considered important with an urgent need to improve the capacity of meteorological and hydrological services to jointly deliver timely and more accurate warnings.

10

APPENDIX: Relevant Recommendations from IWTC-V in 2002 • Storm surge and wave height forecasting is still a significant problem in many tropical

cyclone-affected countries. WMO should endorse and encourage the establishment of storm surge techniques and models, including river flooding and wave action (especially for small islands), for regions that do not have this capability – and furthermore – one of the main impacts of tropical cyclones is the inundation of coastal areas. To address this problem, it is recommended that techniques for forecasting inundation areas be considered and applied, including the use of inundation maps with the combined effects of river floods and storm surge.

• Although scatterometer data were not specifically developed for tropical cyclone

applications, IWTC-V recognises the valuable contribution these data have provided to both the operational and research communities. The meeting recommends that WMO encourage the development of future plans for deployment of scatterometer sensors, and other satellite surface wind vector retrievals within the tropical cyclone.

• Tropical cyclone intensity should not be defined solely by a single parameter such as

central pressure or maximum wind. A more detailed structural analysis is required. The WMO should encourage forecast centres to report within the current WMO format structural information such as quadrant gale radii, eye size, and radius of outer closed isobar in international exchanges of both real-time and best track data. In addition, the time of occurrence and the value of the minimum pressure, and an indication of the occurrence of an eye passage should be added to the synoptic code. This information is needed to determine the maximum wind-minimum central pressure relationship associated with tropical cyclones.

• The meeting endorses the increasing use and application of ensemble prediction systems

(EPS) in forecasting tropical cyclones – and furthermore - it is recognised that ensemble forecasting techniques may provide an important opportunity to improve tropical cyclone track predictions – and furthermore – the meeting recommends that the research community explore the use of ensemble forecasting techniques for tropical cyclone forecasting including track, intensity, Quantitative Precipitation Forecasts (QPF), storm surge, wind waves and flood forecasting.

• The meeting considers small focus workshops to be a useful means to organise

research-operational-hydrological interaction on topics of particular interest to the operational community. The meeting endorses the increasing use and application of ensemble prediction systems (EPS) in forecasting tropical cyclones. WMO should organise a series of thematic workshops on (e.g.,)

o flood forecasting models o forecasting of regional and local flooding due to TCs o storm surge forecasting o extratropical transition of tropical cyclones o ensemble models for QPF of tropical cyclone landfall

• A parametric model of precipitation associated with a landfalling storm should be

developed combining short-range track and intensity forecasts and rainfall rates derived from satellite and radar imagery – calibrated from a rain gauge network.

• Parametric wind models form a basis for a range of forecast and diagnostic applications.

Yet many such models are kept confidential or have not been adequately tested. The meeting recommends that a public domain parametric wind field model, fully tested and verified by peer review, be developed to provide the standard for comparison purposes.

11

• The lack of observational data has caused difficulties with the calibration of the Dvorak intensity analysis technique. It is recommended that calibration or re-calibration of the Dvorak technique, and all pressure-wind relationships, be undertaken in all basins.

• Consensus forecasting techniques have been demonstrated to improve track forecasts

provided an adequate number of skilful forecasts is available. It is recommended that all major NWP centres make available track and intensity forecasts, and radius of gale/storm force winds out to 120 hours or beyond. Consensus forecasts should be closely evaluated to identify the minimum number and optimal combination of forecast members that adds value to the forecast process.

• There is a need for a standard conversion chart that enables users to convert between

different wind-averaging periods and gust factors. The meeting endorses the RSMC recommendation for updating the conversion chart in the Global Guide on Tropical Cyclone Forecasting and requests that the updated values be distributed to TCWCs as soon as they become available.

• The meeting recommends continued development of statistical-dynamical, ensemble

strategies/techniques, dynamical and conceptual models for use in research and real-time operational forecasting of tropical cyclone structure and intensity change over the ocean, during and following landfall – and furthermore – multidisciplinary co-operation is encouraged to address prediction of rainfall associated with tropical cyclones over land at high spatial and temporal resolution and the associated and subsequent effects of flooding, mudslides, and debris flows.

12

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES Topic 0.1: Track forecasts Rapporteur: Lixion A. Avila NOAA/National Hurricane Center 11691 SW 17th Street Miami, FL 33165-2149, USA E-mail: Lixion.A.Avila@noaa.gov Working Group: Phillipe Caroff, Jeff Callaghan, James Franklin, Mark DeMaria Abstract: This paper describes how operational tropical cyclone track forecasting is performed at various centers. The paper also includes track forecast verifications and discusses the issuance and verification of tropical cyclone warnings for the United States. The use of wind speed probabilities by RSMC Miami to better convey forecast uncertainties is introduced. 0.1. Track forecasts The tropical cyclone track forecast is a critical component of the warning system, as it serves as the basis for forecasting the areas threatened by damaging winds, storm surge, and rainfall. RSMC Miami has extended tropical cyclone track forecasts from 3 days out to 5 days, but demand for even longer lead times continues. As with other forecast problems, tropical cyclone prediction relies on a subjective mix of objective guidance with human interpretation and understanding, so that forecaster skill and experience are critical for success. RSMC Miami track forecast errors defined as the distance between a forecast and the subsequently observed position of the storm center averaged 65, 91, 118, 171, 231, and 303 n mi for the 24, 36, 48, 72, 96, and 120 h forecasts, respectively, for the 5-yr period 2001-2005. Using a combination of climatology and persistence as a basis for comparison, track forecast skill exists at all time intervals through 5 days, with the skill increasing out to 36 h and then remaining relatively constant thereafter. The extended 5-day forecasts are currently as accurate as the three-day forecasts were 15 years ago, and forecast errors from 24-72 h are now roughly half of what they were in 1990. During the two very active Atlantic hurricane seasons of 2004 and 2005, RSMC Miami 12-72h track forecast accuracy was at or near record levels. Fig. 0.1.1 shows the long term average track forecast errors for the Atlantic basin.

13

0

100

200

300

400

500

600

700

1954

1957

1960

1963

1966

1969

1972

1975

1978

1981

1984

1987

1990

1993

1996

1999

2002

2005

year

erro

r in

n. m

i.

Tropical Prediction Centeryea rly-average official track forecast errors and trend lines

Atlantic basin, without depressions120-hour 96-hour 72-hour 48-hour 24-hour

The track errors associated with the forecasts issued 24 through 120 hours before a U.S. landfall during the past five years averaged 36, 69, 142, 233, 300 n mi. Figure 0.1.2 shows the average track error for all storms and for those making landfall in the U.S. The track errors for landfalling storms are slightly smaller for the 24 through 72 hours period. There is practically no difference for the 96 through 120-h periods.

Fig. 0.1.1. Yearly official track forecast errors and trend lines for the Atlantic basin.

Fig 0.1.2. Average track errors for storms making landfall in the United States versus all storms.

14

RSMC LA REUNIONTime-evolution of

Cyclone tracks forecasts errors (5-years running means of position errors)(All intensities)

0

50

100

150

200

250

300

350

400

450

90/ 91 -94/ 95

91/ 92 -95/ 96

92/ 93 -96/ 97

93/ 94 -97/ 98

94/ 95 -98/ 99

95/ 96 -99/ 00

96/ 97 -00/ 01

97/ 98 -01/ 02

98/ 99 -02/ 03

99/ 00 -03/ 04

00/ 01 -04/ 05

48-h forecasts

24-h forecasts

12-h forecasts

analysis

Fig. 0.1.4. RSMC La Reunion official average 0-h,12-h, 24-h and 48-h track forecast errors from 1990 through 2005.

The gradual reduction in track forecast errors with time is also evident in other basins. Fig. 0.1.3 shows that for the Australian region the 24-h errors have decreased from 225 km in the early eighties to near 150 km during the past couple of years with a record low of 120 km in 2004. The region issues track forecasts through 24 hours only. Fig. 0.1.3. Official average 0-h, 12-h and 24-h track forecast errors for the Australian region from 1984 through 2005. RSMC La Reunion issues track forecast through 48 hours. Data displayed in Fig. 0.1.4 indicate that there has been no significant change in the 24 hours forecast during the past 20 years. However, a modest improvement has been accomplished at 48 hours. The errors have decreased from 375 km in the early nineties to about 300 km during the past 5 years.

The steady reduction in track error could be attributed to various factors, including the improved quantity and quality of observations from a suite of instruments ranging from satellites to dropsondes, the use of improved data assimilation methods (particularly for unconventional satellite observations).

15

Moreover there has been a significant advance in the resolution and physics of dynamical forecast models. Fig. 0.1.5 shows the gradual reduction in the 48-h track model errors.

Fig. 0.1.5. 48-h track forecast errors from 1965 to 2005.

Despite the improvements in track forecast accuracy, forecast uncertainty requires the issuance of U.S. hurricane warnings for relatively large coastal areas. During the period 2000-2005, the average length of a hurricane warning in the United States was 275 n mi. This represents a decrease from the preceding decade, during which the average warning length was 395 n mi, and appears to represent a reversal of an earlier trend toward larger warning areas. Even so, only about one-quarter of an average hurricane warning area experiences hurricane conditions. Another important factor is the lead-time of hurricane warnings. This is defined as the time between the issuance of the warning and the time of experiencing hurricane force winds at a warned point.

The average lead-time of NHC hurricane warnings has increased from 19 hours in the seventies to near 34 hours for the period 2000-2005. See Figure 0.1.6.

16

Fig. 0.1.6. Average lead-time of hurricane warnings for the United States.

Continued improvements in track forecast accuracy could be used to decrease the amount of “over-warning”. However, hurricane warnings must carefully balance the need to minimize the negative impacts of over-warning against the need to safeguard lives and property. The continual increase in coastal population and development without concurrent expansion of evacuation routes has resulted in increased evacuation clearance times. As a result, the NHC has put greater attention on translating forecast improvements into increasing the lead-time of warnings, instead of decreasing the warning area. Hurricane Katrina reminds us that in spite of all the progress that has been made in forecasting the path of tropical cyclones, the United States remains vulnerable to a large loss of life from storm surge. RSMC Miami has been using a cone of uncertainty to display graphically the uncertainties of a track forecast as shown in Fig 0.1.7. This cone is constructed using the past forecast track errors.

Fig. 0.1.7. Cone of uncertainty associated with Hurricane Katrina, 2005.

17

Experimental TPC WebsiteExperimental TPC WebsiteWind Speed Probability GraphicsWind Speed Probability Graphics

34 kt 34 kt exampleexample

Rita Rita advisory advisory

#17#17

RSMC Miami has also begun to use “wind speed probabilities” to better convey the uncertainty in forecast. This new product is about a weather event at any specific location and conveys chances of wind speeds of at least particular thresholds 34- and 50 -kt (tropical storm force) and 64 kt (hurricane force) and in fact, it accounts for combined uncertainty in track, intensity, and size. This method is based on a large set of plausible tracks and intensities (ensemble members) roughly centered around the current official forecast and determined by random sampling of historical track and intensity errors in official forecasts (since 2001). Fig 0.1.8 shows the wind speed probability associated with Hurricane Rita 2005.

Fig. 0.1.8 Probability of 34-kt winds associated with Hurricane Rita, 2005. 0.1.1 Practical operational remarks In making the track forecast, it is suggested to maintain continuity with previous forecast. It is better to have a modest response lag than to jerk the forecast back and forth. The official forecast rarely deviates from the guidance envelope and the guidance spread is in general reflected in the forecast confidence. For example, a large model spread as shown in Fig. 0.1.9 implies low confidence in the forecast. The opposite is not necessarily true. A multi-model consensus is a very powerful tool for producing accurate track forecast. However, it is important to examine the synoptic data and the model fields and do not focus in the predicted TC track in the model. This is also important in identifying a model outlier but this is very difficult.

18

Fig. 0.1.9. Track guidance associated with Hurricane Ophelia at 1200 UTC 9 September. Note the large spread in the models implying low confidence in the forecast. Acknowledgments: This section was prepared with the information provided by Mark DeMaria, Jeff Callaghan, Phillipe Caroff, James Franklin and presentations prepared by the Hurricane Specialist Unit at RSMC Miami.

19

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES Topic 0.2: Observations and Forecasts of Wind Distribution Rapporteur: S.T. Chan

Hong Kong Observatory 134A, Nathan Road, Kowloon, Hong Kong, China

E-mail: stchan@hko.gov.hk Fax: +852-2375-2645 Working Group: Kevin Cheung, Akhilesh Gupta, Bruce Harper, Jeffrey D. Kepert, Kenichi Kusunoki 0.2.1 Introduction Forecasting of wind distribution of a landfalling Tropical Cyclone (TC) is definitely an operational challenge. It is of great importance since the circulation of a TC making landfall would inflict huge losses and widespread damage on the coastal areas. As outlined in Willoughby et al. (2005a), one of the priorities of US Weather Research Program on TC is to make skillful forecasts of gale- and hurricane-force wind radii out to 48 hours with 95% confidence. While there has been much improvement over the years in the forecasting of TC tracks, relatively slow progress was seen during the same period on the problem in question, primarily due to the complexities in the physical processes involved, and the inability of Numerical Weather Prediction (NWP) models to be run at a resolution high enough to adequately resolve the TC structure in an operational manner. Only a limited number of objective guidance tools have been specifically designed for the task and statistical methodologies remain to be the mainstay in tackling the problem. In this report, new developments since IWTC-V are reviewed. The recent researches on TC structural and wind distribution evolution arising from land-sea contrasts, in particular the influence of land-induced asymmetric friction on the boundary layer winds, and the effects of the moist processes in introducing asymmetries in TC structure at landfall, is covered in Sections 0.2.2a and 0.2.2b. Damaging fine-scale surface wind features such as boundary-layer rolls, small-scale spiral bands, terrain-induced wind accelerations are frequently observed. A brief account of these features is given in Section 0.2.3a. Observational issues including a review of the wind-pressure relationship, deployment of surface wind observation network and the emergence of new wind speed averaging standards are discussed in Sections 0.2.3b, 0.2.3c and 0.2.3d. On the forecasting aspects, the development and operational deployment of various empirical and parametric wind models are reviewed in Sections 0.2.4a and 0.2.4b. Latest progress in the modeling of the boundary layer processes in NWP is summarized in Section 0.2.4c. 0.2.2 Recent Understanding of TC Structural Change due to Land-Sea Contrasts The wind structure of a mature TC is basically axisymmetric. However, when a TC makes landfall, part of the TC circulation is affected by increased surface friction over land. In addition, the surface sensible and latent heat fluxes, and moisture supply over land are different from those over ocean. The major goals of recent researches on TC landfall have been to understand how the contrasts between

20

land and ocean will affect the various physical processes, thereby introducing asymmetries in the TC structure.

a) Land-induced Asymmetric Friction The increased friction over land has long been recognized as an important influence on the wind structure changes that occur at landfall (for example, Powell 1987). More recently, it has been recognized that the landfall-induced asymmetric friction provides a predominantly wave-number one forcing, as does the motion-induced asymmetry. Thus the asymmetric boundary-layer wind structure in a stationary storm partly over land should be similar to that in a moving storm. Blackwell (2000) presented an observational analysis of the flow in Hurricane Danny while it was nearly stationary at landfall on the US Gulf Coast, which showed a marked wind asymmetry, with a 41 m s-1 maximum at about 500 m altitude in the offshore flow, in contrast to a 31 m s-1 maximum at 1500 m in the onshore flow. Kepert (2002a) argued that this asymmetry was similar to those in 3-dimensional models of a moving storm (Kepert 2001, Kepert and Wang 2001), and that the only essential difference was the source of the asymmetric frictional forcing: motion, or proximity to land. He presented model results, using the 3-dimensional boundary layer model of Kepert and Wang (2001), which were in excellent agreement with Blackwell’s observations. A subsequent study (Kepert 2002b) investigated the response when a moving storm makes landfall, that is, both sources of asymmetric forcing are present. It was found that as the storm makes landfall, the motion-induced wind maximum in the right forward quadrant weakens, while a secondary maximum appears in the offshore flow to the left of the track (in the Northern Hemisphere). Shortly before landfall, both are present, and the modeled surface wind field is strikingly similar to Powell’s (1980) analysis of the landfall of Hurricane Frederic. In a landfalling cyclone, the wind speed over land is reduced less by friction than would be the case for straight flow, as the resulting enhanced inflow gives increased angular momentum advection which helps to maintain the azimuthal flow component. When the storm is near land, this strong inflow extends over the sea on the offshore-flow side of the storm, due to advection of this inflow by the swirling flow, and causes particularly strong surface winds in the offshore flow. Kepert (2002b) also included a comparison of modeled and GPS-dropsonde observed wind profiles in Hurricane Floyd, with strong agreement. Schneider and Barnes (2005) analysed the landfall of Hurricane Bonnie, and similarly found a region of enhanced near-surface inflow to the southwest of the storm in the offshore flow (that is, to the left of the track). They argued that the unusual location here is consistent with greater frictional forcing over the land giving an enhanced cross-isobar angle. They noted also that this inflow, which entered the southern part of the eyewall, was relatively cool, dry and stable, consistent with its over-land origin. Hurricane Mitch made an extraordinarily slow landfall upon Honduras, taking some 36 hours to travel the last 100 km. Kepert (2006b) analysed dropsonde data from a period when the storm was about 80 km off the coast and moving towards it at about 2 m s-1. The flow showed most of the features that would have been expected from the motion-induced frictional asymmetry. In particular, the position of the strongest storm-relative winds rotated anticyclonically with height, and the strongest inflow was always about 90o of azimuth upstream of the strongest azimuthal winds and similarly rotated anticyclonically with height. However, the surface wind maximum was located to the left rear of the storm, rather than in the right front as would be expected if motion was forcing the asymmetry. Modelling results were presented which showed that land at roughly three times the RMW was able to produce a flow asymmetry that propagated inwards with the boundary-layer inflow to produce a marked asymmetry at the RMW.

21

Figure 0.2.1: Objective analyses of the storm relative (top) azimuthal and (bottom) radial wind components of Hurricane Mitch, for levels as shown, based on dropsonde data. Contour interval is 5 m s-1, with multiples of 20 m s-1 shown heavy. Darker shading corresponds to stronger azimuthal wind and stronger inflow, respectively (Kepert, 2006b).

As well as inducing a marked asymmetry in the inner core, friction due to proximity to land also produces larger-scale asymmetries in the storm. Wong and Chan (2006a) analysed the structure of the friction-induced surface convergence asymmetry, showing that this would be largest at the coast on the offshore-flow side of the storm in the core, but on the onshore-flow side at larger radii, which could be understood in terms of the storm-scale response to the asymmetric forcing. In another paper, Wong and Chan (2006b) studied the effect of this asymmetry on the storm motion, and found that it was capable of producing a landwards drift of ~1 m s-1 when the storm is 150 km offshore. They analysed the cause of this drift, and found that it was not primarily due to the asymmetric flow, but that generation of potential vorticity by the asymmetric vertical motion and diabatic heating was also important. The asymmetric vertical motion was produced partly as a response of the vortex to vertical tilt of its axis, and partly by asymmetric boundary layer convergence. This kind of effect on storm motion is important in considering the rate of increase in wind magnitude when a TC is approaching the coastline.

Shum and Chan (2006) extended this study to the case of a moving storm. The surface wind structure was in good agreement with that described by Kepert (2002b), but their study also analysed the vertical motion and rainfall. They find a marked tendency for the updraft at the top of the boundary layer to lie to the right front of the storm, and strengthen after landfall. The rainfall field is significantly less asymmetric than the updraft at the top of the boundary layer, and displays a distinct cyclonically-propagating maximum after landfall. They also observe an increase in the low-level tilt of the vortex after landfall, and argue that this contributes to the storm weakening, as well as the reduction in the surface thermodynamic fluxes at landfall.

22

b) Moisture Supply and Thermodynamic Flux Changes Chan and Liang (2003) conducted simulations of the landfall process of a vortex on an f-plane using the fifth-generation Pennsylvania State University-National Center for Atmospheric Research Mesoscale Model (MM5). Sensitivity experiments were performed by applying different land conditions: no sensible heat flux, no moisture flux, a higher surface roughness, or moisture supply limited to the lowest level over ocean. The results show that the change in sensible heat flux has little effect in modifying the convective structure of the TC, and that the moisture flux is the dominant factor. Due to the latter, maximum precipitation is found to the front and left quadrant (with respect to the landfall direction) of the TC, which is consistent with observations (Chan et al. 2004). However, the low-level convergence along the coast line does not significantly change the inner structure of the TC in the simulations. Rather, the overall stability (and thus the convection distribution) is modified when the dry air over land is advected by the TC primary circulation. Chen and Yau (2003) also performed MM5 simulations of an idealized landfall process but with higher resolution (6 km vs. 15 km) so that moist processes can be simulated explicitly. Then potential vorticity (PV) and Eliassen-Palm (EP) flux analyses were utilized to diagnose the physical processes during landfall. Probably due to the higher resolution in the Chen and Yau simulations, more storm-scale features were identified. There is a band of positive PV ahead of the TC that develops along the coastline (Fig. 0.2.2), and the interaction of this PV band with the eyewall PV ring leads to fluctuation in the intensity. The authors suggested that this type of interaction in the boundary layer could be responsible for some eyewall replacement cycles. In fact, an eyewall contraction, breakdown and reformation process was observed in a typhoon during landfall (Wu et al. 2003). Another important conclusion from the Chen and Yau study is that the effect of diabatic heating was found to be quite important in the spinning down of the TC vortex during landfall. Computations show that the contribution at the lower levels to the tangential wind change before landfall is ~30 m s-1 h-1 by diabatic heating and only ~10 m s-1 h-1 by the eddies.

Figure 0.2.2: Model simulated evolution of the horizontal structures of PV (darker more positive). The thick lines denote the shore (Chen and Yau, 2003).

0.2.3 Observational Aspects

a) Fine-scale Surface Wind Structure Recent significant progress has been made in understanding the dynamics of small-scale features in the boundary layer winds, on both the observational and theoretical fronts. One area of difference

23

between these has been that the observational studies have necessarily focused on storms near land, while the theory has so far ignored the role of landfall on these processes. From the observational studies, one could perhaps draw the incorrect conclusion that these small-scale features are only present near land, but it would be more correct to say that it is near land that the observations available with sufficient spatial and temporal resolution to capture these small-scale features, while theory suggest that they should be ubiquitous. The dynamics of these fine-scale features are considered in more detail in Topic 1.2. Boundary-layer rolls are very common in the atmospheric boundary layer (see e.g. Etling and Brown 1993 for a review). Wind circulations associated with these rolls may produce highly organized and damaging surface winds (Wakimoto and Black 1994). Wurman and Winslow (1998) presented the first Doppler radar evidence for their existence in tropical cyclones, indicating intense horizontal roll vortices with an average wavelength of 600 m roughly aligned with the mean azimuthal wind. Several papers have since presented similar evidence. Katsaros et al. (2002) examined SAR images of Hurricanes Mitch and Floyd and also found periodic kilometer-scale variation. More recently, Morrison et al. (2005) describe features that are significantly less streaky in appearance, to the extent that it is not entirely clear that they are the same phenomenon. Possibly the different radar technology used by the groups may have contributed to this difference, since Morrison et al. (2005) also show a SAR image which displays parallel streaks more similar to classical boundary layer rolls. In Japan, a new ground-based observational study of typhoons using Doppler radar for Airport Weather (DRAW) has been conducted from 2005 (Kusunoki and Mashiko 2006; Kusunoki 2006). The preliminary analysis in the inner core of Typhoon Songda (2004) before and during landfall on the Okinawa Island indicates that the perturbation reflectivity field has many small-scale spiral structures spiraling outward from the eyewall (Figure 0.2.3), which are approximately similar to those shown by Gall et al. (1998), Wurman and Winslow (1998), and Morrison et al. (2005). Lorsolo et al. (2006) and Wurman et al. (2006) analyse the vertical structure of the rolls, and find them to be coherent through the depth of the boundary layer (~500 m), and compare radar- and tower-measured winds with fair agreement, demonstrating that the roll circulation extends to the surface, albeit with attenuation and other scales of motion superimposed.

Dis

tanc

e fro

m th

e ra

dar (

km)

dBZ

Dis

tanc

e fro

m th

e ra

dar (

km)

dBZ

Figure 0.2.3: Perturbation reflectivity field for the inner-core region of Songda taken at 1203 JST 5 September 2005 (Kusunoki and Mashiko, 2006). Theoretical analyses of roll development in tropical cyclones were provided by Foster (2005) and Nolan (2005). Foster (2005) argues that the tropical cyclone boundary layer is an ideal environment for roll development. His argument extends the classical theory of roll development as an inflection-point instability of the frictionally-induced cross-isobar flow to the case of a tropical cyclone. Here, the cross-stream shear and hence instability are strong because the boundary layer is relatively shallow,

24

and the cross-stream component in analytical solutions is stronger than in classical Ekman-like solutions for straight flow (Kepert 2001). Nolan (2005) presents a stability analysis of a symmetric vortex, and finds both symmetric and asymmetric responses. The instabilities acquire some energy from the shear in the radial flow near the top of the boundary layer, in which regard they are similar to Foster’s (2005) rolls. However, Nolan (2005) shows that the vertical shear of the azimuthal wind can also contribute energy to the instability, and that the relative importance of these mechanisms depends on the inertial stability of the storm and on the orientation of the mode. The GPS dropsonde has now been operational for close to a decade, and several thousands have been deployed in the eyewall of hurricanes. Extreme gusts have been reported in both horizontal and vertical wind components (Aberson and Stern 2006, Henning 2006, Stern and Aberson 2006). Such extreme events contribute greatly to the extensive wind damage brought by a landfalling TC that could not be adequately represented by broad-brush destructive potential scales such as the Saffir Simpson Scale. While the extreme gusts are (by definition) rare events, the steadily increasing sample is beginning to enable statistical characterization of their nature. Another type of fine-scale feature that is definitely unique to landfall is topographically-induced accelerations. These may take the form of shear lines, reversed flow, small-scale vortices, streaks, and downslope winds, due to the complex interaction of topography and flow. Shun et al. (2003) presented interesting Doppler radar observations of some of these phenomena in Hong Kong, and well illustrated the range and complexity of the problem. Mueller et al. (2006) compared exposure-based engineering models with observed damage on Bermuda during Hurricane Fabian. This is important work, as such models are foundational to the design of structures and to climatological risk analysis, but have not been extensively verified in tropical cyclones. Similar analysis is being undertaken in Australia following the extremely damaging landfall of Severe Tropical Cyclone Larry in northern Queensland in 2006.

Figure 0.2.4: Doppler radar observation of high speed streaks (positions M-I, M-II, and M-III) and vortex (position A) on 7 June 1999 as Typhoon Maggie was making landfall some 150 km west of Hong Kong. White arrow indicates the background flow direction (details in Shun et. al, 2003).

b) Wind-Pressure Relationship Harper (2002) provided an overview of the historical development of wind-pressure relationships and their application within the Australian region, highlighting many aspects that affect the quality of the best track databases throughout the world, and challenging the justification for using basin-specific wind-pressure curves. Harper argues that the variability in wind-pressure balance within and between

25

individual storms is likely much greater than any regional “average” and that it is important to incorporate storm spatial scale and profile shape to allow for the expected natural variability in the wind-pressure balance. These conclusions supported, in principle, the operational experience already documented by, for example, Callaghan and Smith (1998) and Guard and Lander (1996) on how to characterise “midget” systems. In particular, Harper retraced the early development of the Dvorak technique to show that the current Atlantic wind-pressure relationship was actually developed using mainly NWP storms but was dropped in preference to the Atkinson and Holliday (1977) results. Aspects of this investigation have also been reported in Velden et al. (2006). Harper also argued for reconsideration of the Holland (1980) B parameter as an indicator of the wind pressure variability. Recently, Knaff and Zehr (2006) have further examined the basis of the Atkinson and Holliday (1977) relationship and provide a modified best-fit to the original dataset, which yields a result closer to the Dvorak Atlantic curve. Their method also provides a basis for utilising the readily available operational parameters of size, latitude and environmental pressure to enable forecasters to adopt specific wind-pressure pairings on a case by case basis. The authors claim that using the proposed unifying equations, the MSLP can be estimated from the Vmax within 5 to 6 hPa and the wind can be estimated from the MSLP within 7 to 8 kt. Recently Weber (2006) has also proposed an alternative wind profile approach that uses radius of outermost close isobar and central and environmental pressure. As noted by Knaff and Zehr (2006), improved accuracy in estimating central pressure will also have value in the better initialization of NWP models.

c) Surface Wind Observation Network Notwithstanding the increasing importance and utility of remote sensing systems (e.g. satellite MI proxy and scatterometer) for estimating surface wind speeds, there remains a critical need to maintain and expand surface wind measurement networks in all areas where communities are at risk from tropical cyclone impacts. Verification of forecast/modelled winds must remain a priority if real improvements in techniques and procedures are to be realized. There are several challenges in this regard, the first being the costs of establishing and maintaining instrument systems. Others include the need to optimally locate such instrumentation or in lieu, to ensure that less than optimum sites are adequately calibrated. Finally, baseline instrumentation needs to be ruggedised and possess backup power and data storage to enable full recovery of information from extreme wind events (recent instrument failures of note include TC’s Ingrid and Monica across Northern Australia and Hurricane Katrina on the US Gulf Coast). A critical development in recent years, mainly in the US, has been the availability of mobile wind instrumentation systems, typically owned and operated by research organizations (e.g. Schroeder and Smith 2003). These systems are now highly developed and their use has added an enormous amount of knowledge to the science of the near-surface land boundary layer under tropical cyclone conditions. The impetus and initiative for these systems has come mainly from a wind engineering focus but there should be no reason why such mobile systems could not be usefully deployed as a part of a comprehensive forecast and verification system by mainstream meteorological agencies. Opportunities for utilising new technologies should also be considered. For example, the proliferation of wireless networking protocols and low-cost low-power electronics continues to widen the opportunities for “smarter” wind sensing systems, potentially delivering a much greater density of measurements at a lower cost. The use of “infrastructure of opportunity” rather than relying on baseline installation of standard height towers also has the capacity to reduce capital costs. Such infrastructure can include power transmission line towers, communications towers and the like. Typically, operators of such equipment also have a vested interest in having access to long term wind measurements (e.g. for power transmission efficiencies or damage assessment etc) and symbiotic relationships and partnerships may well develop. Cheaper and even disposable (nil maintenance or “bolt-and-forget”) wind sensor systems should be possible with existing technologies. Importantly though, the adopted infrastructure site must be calibrated so that the attached sensor delivers a known directional response

26

that can be reliably converted to standard exposure and directly ingested by forecast and/or modelling systems. Standardisation of wind sampling and analysis is essential so that data collected from one environment can be reliably compared with another and work to build a reliable ground truth, from which better models and procedures can emerge. Without attention to these basic needs, the still significant “noise level” in wind forecast capabilities will never be reduced. Powell et al. (2004) presented a project in the United States to photographically document exposures of over 200 automatic weather stations in hurricane-susceptible areas. Roughness lengths for each octant of wind direction for all documented stations were estimated, from which the wind measurements can be corrected to open terrain for use in real-time analyses of hurricane wind fields. The exercise demonstrates that mean wind measurements associated with significant upstream terrain may underestimate the open-terrain wind by about 30%.

d) Wind Speed Averaging Standards One of the difficulties in transferring forecast techniques from one region to another has been the use of different “standards” for reporting the wind strength. While the WMO standard remains as the 10 minute average wind for synoptic reporting, adopted by RA I and RA V, other approaches have developed over time due to local preferences. RA IV (Americas and Caribbean) have adopted a 1-minute average (termed “sustained”) for all warning purposes. The remaining regions (ESCAP TC and Typhoon) use the 10-minute average but allow a 3-minute average for non-recording observations and, in China, a 2-minute average is recognised. Paradoxically, none of the WMO associations define a “gust” wind standard, although it is recognized that short period wind gusts (say 2- to 3-second) are responsible for the greater proportion of community damage and are typically the averaging values used by planning and standards bodies that oversee the mitigation of tropical cyclone threats through improved building codes and the like. Additionally, the adoption of a 1-minute “sustained” wind by RA IV, US territories in the Western North Pacific and the US military globally, has led to the 1-minute wind becoming the de facto standard for the application of the Dvorak (1984) technique, with other regions applying conversions (e.g. WMO 1993, Table 4.2) to suit their local needs. Unfortunately, there is evidence that the need for such conversions may have contributed to the introduction of systematic errors in some best track archives (e.g. Harper and Callaghan 2006). Also, it is typically not appreciated that the Dvorak “Atlantic” wind-pressure relationship has no stated wind averaging context but was simply the maximum expected surface wind (Velden et al. 2006) and, by association with the wind and pressure curve has been interpreted as 1 minute. These differences and/or misunderstandings have allowed a situation to develop over many decades whereby some potentially avoidable variance (of the order of 20%) in wind measurements and/or estimation has entered the science, notwithstanding that TC intensity estimation is an already extremely difficult problem. Following IWTC V in Cairns in 2002, the WMO acted to reduce the uncertainty in converting between the various wind averaging contexts by commissioning a best practice review (Harper et al. 2004), which is expected to be finalized soon. The review considers the theoretical turbulence framework within which wind-averaging conversions are valid (or indeed invalid) and also extensively reviews both historical data (especially TC data) and theoretical statistical models for estimating such conversions. A revised set of conversion factors is planned that will supercede that presently in WMO (1993) and provide guidance on when and how to apply such conversions in both forecast and research environments.

27

0.2.4 Forecasting Aspects

a) Empirical Models of Wind Distribution for Landfalling TCs For operational early warning and disaster mitigation purposes, some empirical models of wind distribution for landfalling TCs are currently applied. Some of these models derive empirical laws of the weakening of a TC during landfall based on historical cases (e.g., Vickery 2005 for the Gulf of Mexico Coast and the coast of the Florida Peninsula). In some others, the maximum sustained surface wind of a TC is assumed to decay exponentially during landfall (e.g., Roy Bhowmik et al. 2005 for the east coast of India). However, note that the mentioned applications of empirical models are for relatively flat coastal areas. When orographic influence is effective, the TC track may be deflected and the wind distribution in the landfall area will be much different from that without topography (e.g., Lin et al. 2005).

b) Parametric Wind Field Modelling This is an area that remains potentially under-utilized and under-valued by many forecast centers, where the traditional use of Dvorak (1984) has perhaps been seen as the only legitimate technique for estimating intensity when aerial reconnaissance is not available. For example, in IWTC IV Topic 4.3 (WMO 1998) a number of parametric modelling tools were presented, typically based on the Holland (1980) or similar technique, or even involving the conceptual Rankin vortex etc. Such techniques do rely on having some relevant surface wind (and/or pressure) observations but can be readily applied within a simplified modelling context to add significant value to satellite-only intensity estimates. The parametric model has the advantage of providing a stable theoretical quasi-static force balance which, with the benefit of time-history and spatial contexts, can readily augment the Dvorak analysis, which has neither of these attributes. The operation of parametric models has been boosted in recent times by the ready availability of sea surface scatterometer data, which can now very effectively provide the outer spatial scale of storm systems (e.g. Rgales). When combined with an independent estimate of the inner scale (Rmax), typically via satellite (VIS or EIR or MI), a full parametric wind field can be readily estimated. The emerging EIR method by Kossin et al. (2006) promises to also provide both of these parameters as well as estimates on the entire two-dimensional surface wind field within 200 km of the storm centre in an automated fashion. With such objective data, parametric models could also include asymmetries and conservative properties such as angular momentum can also be considered. Parametric models have the capability of providing a significantly similar quality of surface wind estimates to full NWP models but at a fraction of the cost and effort. Such models could readily be used to replace what are still typically hand-drawn warning and evacuation zone products at many centers. A potentially significant advance in parametric modelling was recently made available through the work of Willoughby et al. (2005b), whereby several decades of US aerial reconnaissance flights were analysed to obtain a family of radial gradient windspeed profiles. This analysis represents the most significant advance yet in describing the radial wind field of tropical cyclones and has additionally provided very valuable information on data dependencies of spatial scale, intensity and latitudinal variations, at least within the Atlantic basin. Other techniques that remain promising include the “double Holland” concept introduced by Thompson and Cardone (1996), which can be used to represent eyewall replacement cycles, and the more sophisticated yet efficient analytical Ekman modelling approach as presented by Kepert (2001), which provides for radially and azimuthally varying inflow and gradient height reduction, albeit within the assumption of an applied Holland gradient level pressure profile. SHIPS (Statistical Hurricane Intensity Prediction Scheme) is a successful statistical-dynamical model deployed by NHC for operational intensity forecasting in the Atlantic and East Pacific basins. The scheme involves multiple linear regression of environmental and “CLIPER”-type parameters. A method to adjust the real-time forecasts over land using a simple empirical exponential decay model

28

has since been introduced in 2000 to the scheme (DeMaria et al., 2005). The inclusion of the effects of the decay over land reduced short-range Atlantic and East Pacific intensity errors up to 72 hours. DeMaria et al. (2006) further presented a modified decay model for storms that move over narrow landmasses. The modified decay model includes a factor equal to the fraction of the storm circulation that is over land and it was applied to SHIPS in 2005. The new scheme reduced the intensity forecast errors by up to 8% relative to the original decay model for cases from 2001 to 2004 in which the storm was within 500 km from land (Figure 0.2.5). The version of SHIPS for the western North Pacific basin - Statistical Typhoon Intensity Prediction Scheme (STIPS), was developed and implemented operationally at the Joint Typhoon Warning Center (JTWC) in July 2002 and updated in mid June 2003 (Knaff et al., 2004).

Figure 0.2.5: Percentage improvement in the mean absolute intensity error of the 2001-04 operational SHIPS forecasts when the modified decay model is introduced. Results are shown for the total sample and the sample in which the best track was within 500 km of land (DeMaria et al., 2006) Several modifications are being planned for SHIPS, which include predictors from aircraft flight-level observations (with inner-core wind field structure information), daily SST analyses from Advanced Microwave Scanning Radiometer for the Earth-Observing System (AMSR-E) and total precipitable water analysis from microwave imagery.

c) Modelling of Boundary Layer Processes in NWP As seen from the discussions on theoretical development, the evolution of the intensity and structure of a TC has strong dependence on changes in the surface conditions. Successful prediction of structural changes of a landfalling TC depends much on the adequacy of our current NWP models in simulating the boundary layer processes (e.g., Shen et al. 2002). So far, little has been known about the exchange and drag coefficients at high wind conditions largely due to the fact that measurements under extremely high wind conditions are very difficult to make. Based on GPS dropsonde measurements, Powell et al. (2003) showed that as surface winds increase above 40 m s-1, the sea becomes completely covered by a layer of foam which impeded the transfer of momentum form the wind to the ocean. As a result, drag coefficients would decrease with the wind speed. This finding helps reduce the uncertainties in the calculation of surface fluxes and thus improving TC intensity forecast by NWP models. In their review, Wang and Wu (2004) called for a prompt development of a new parametrization scheme for the drag coefficient based on Powell et al. (2003) and an evaluation of its effect on TC intensity.

29

Kasheta and Chang (2002), based on a “dry” hurricane boundary-layer model, demonstrated that the downward transfer of high momentum aloft played a significant role in the maintenance of high wind speeds at the surface with the surface wind maxima being observed on the lee sides of high terrain. They suggested that a refined surface roughness length scheme over land should be considered for real terrain studies, e.g. a scheme based on the surface canopy as well as terrain height. Such a scheme in conjunction with a very fine grid (say 1 km mesh) would allow the microscale structure of terrain-induced downdrafts to be captured. Recently, Wu et al. (2003) studied the eyewall evolution of Typhoon Zeb (1998) and found that the eyewall evolution comprised an eyewall contraction just before landfall at Luzon, followed by a breakdown after landfall and then a reformation of the eyewall upon reentering the ocean as the storm left Luzon. They showed that a high-resolution model could reproduce such an eyewall evolution but the physical processes responsible for that are not yet understood. Further investigations into such eyewall evolution processes would improve our understanding of TC structure and intensity changes for TCs making landfall and interacting with mesoscale terrains (Wang and Wu, 2004). In the coming years, studies on sensitivities to various parameters in a land surface model when simulating a TC landfall and evaluation of the skill of the NWP models in forecasting the associated wind distribution are needed (e.g., Farfán and Cortez 2005; Rodell et al. 2005). 0.2.5 Summary and Recommendations A summary of possible research efforts and recommendations about future research directions in dealing with the roadblocks and requirements discussed in the previous sections is given below. 1. It is expected that the motion and landfall contributions to asymmetric friction will more commonly be of comparable magnitude. The interaction between them and their impact on TC structural evolution may be important and full investigation of this topic is needed. 2. Extreme wind gusts observed in landfalling TC, whether induced by convective, coherent (e.g. rolls) or vortex-related structural features could be responsible for increased damage at ground level. These are not necessarily represented by the broad-brush destructive potential scales currently used. With the accumulation of more observational data, by combining GPS dropsondes, Doppler radar and fine scale tower surface wind measurements, better characterization of such destructive phenomenon should become possible. 3. The recent important work using an exposure-based engineering model to quantify the impact of topographic speed-up effects on the observed damage to structures during Hurricane Fabian in Bermuda was noted. Such models are foundational to the design of structures and to climatological risk analysis, and should be extensively verified in tropical cyclones. 4. Parametric wind field models could usefully augment the Dvorak analysis to provide quality surface wind estimates at a fraction of the cost and effort required for NWP modelling. More extensive use of them by operational forecast centres is recommended. 5. A refined surface roughness length scheme over land should be considered for real terrain studies. Such a scheme in conjunction with a very fine grid would be able to capture the microscale structure of terrain-induced downdrafts. 6. Studies on sensitivities to various parameters in a land surface model when simulating a TC landfall and evaluation of the skill of the NWP models in forecasting the associated wind distribution are needed. In particular, progress has been made in understanding the air-sea exchange under high

30

wind speeds, efforts in similar validation of the land surface schemes are also required. 7. Verification of forecast/modelled winds must remain a priority if real improvements in techniques and procedures are to be realized. Agencies should consider the need for enhancing and expanding surface wind networks in tropical cyclone prone areas and investigate more innovative alternatives to conventional fixed multi-parameter weather stations so as to improve the chances of obtaining verifying surface wind data. Wind measurement and reporting standards must also be improved to ensure consistency across the various forecast techniques and NWP estimates. Bibliography

Aberson, S. D., and D. P. Stern, 2006: Extreme horizontal winds measured by dropwindsondes in hurricanes. Extended abstracts, 27th Conference on Hurricanes and Tropical Meteorology. Amer. Meteorol. Soc., Monterey, CA, April 24 – 28. Paper 16B7. Atkinson G. D., and C. R. Holliday, 1977: Tropical cyclone minimum sea level pressure/maximum sustained wind relationship for the Western North Pacific. Mon. Wea. Rev., 105, 421-427. Bessho K., M. DeMaria and J. A. Knaff. 2006: Tropical Cyclone Wind Retrievals from the Advanced Microwave Sounding Unit: Application to Surface Wind Analysis. J. Appl. Met. and Clim., 45, 399-415. Blackwell K.G., 2000: The evolution of Hurricane Danny (1997) at landfall: Doppler-observed eyewall replacement, vortex contraction/intensification, and low-level wind maxima. Mon. Wea. Rev.128, 4002-4016. Callaghan J. and R. K. Smith, 1998: The relationship between surface wind speeds and central pressure in tropical cyclones. Aust Meteorological Magazine, 47, 191-202. Campo, K. D. and T. M. Riskenback, 2006: Eyewall evolution of Hurricane Katrina newr landfall using NEXRAD reflectivity and radial velocity data. Extended abstracts, 27th Conference on Hurricanes and Tropical Meteorology. Amer. Meteorol. Soc., Monterey, CA, April 24 - 28. Paper P7.10 Chan, J. C. L., and X. Liang, 2003: Convective asymmetries associated with tropical cyclone landfall. Part I: f-plane simulations. J. Atmos. Phys., 60, 1560-1576. Chan, J. C. L., K. S. Liu, S. E. Ching, and E. S. T. Lai, 2004: Asymmetric distribution of convection associated with tropical cyclones making landfall along the South China coast. Mon. Wea. Rev., 132, 2410-2420. Chen, Y., and M. K. Yau, 2003: Asymmetric structures in a simulated landfall hurricane. J. Atmos. Phys., 60, 2294-2312. Cubukcu, N., R. L. Pfeffer, D. E. Dietrich, 2000: Simulation of the effects of bathymetry and land-sea contrasts on hurricane development using a coupled ocean-atmosphere model. J. Atmos. Sci. 57, 481-492. DeMaria M., M. Mainelli, L. Shay, J. Knaff and J. Kaplan, 2005: Further Improvement to the Statistical Hurricane Intensity Prediction Scheme (SHIPS). Wea. Forecasting, 20, 531-543. DeMaria M., J. A. Knaff and J. Kaplan, 2006: On the Decay of Tropical Cyclone Winds Crossing Narrow Landmasses. J. Appl. Met. and Clim., 45, 491-499.

31

Dvorak V. F., 1984: Tropical cyclone intensity analysis using satellite data. NOAA Technical Report NESDIS 11, 45pp. Eckman, R. M., R. J. Dobosy, T. W. Strong and P. G. Hall, 2006: In situ measurements of 3d turbulence in Hurricanes Frances and Ivan using a pressure-sphere anemometer. Extended abstracts, 27th Conference on Hurricanes and Tropical Meteorology. Amer. Meteorol. Soc., Monterey, CA, April 24 - 28. Paper 10C4. Farfán, L. M., and J. A. Zehnder, 2001: An Analysis of the Landfall of Hurricane Nora (1997). Mon. Wea. Rev., 129, 2073-2088. Farfán, L. M., and M. Cortez, 2005: An observational and modeling analysis of the landfall of Hurricane Marty (2003) in Baja California, Mexico. Mon. Wea. Rev., 133, 2069-2090. Foster, R. C., 2005: Why rolls are prevalent in the hurricane boundary layer. J. Atmos. Sci., 62, 2647-2661. Gall, R., J. Tuttle, and P. Hildebrand., 1998: Small-scale spiral bands observed in Hurricanes Andrew, Hugo, and Erin. Mon. Wea. Rev., 126, 1749-1766. Geerts B., Heymsfield G. M., Tian L., Halverson J. B., Guillory A., Mejia M. I., 2000: Hurricane Georges's landfall in the Dominican Republic: Detailed airborne doppler radar imagery. Bull. Amer. Meteor. Soc., 81, 999-1018. Guard C. P. and M. A. Lander, 1996: A wind-pressure relationship for midget TCs in the Western North Pacific. In: 1996 Annual Tropical Cyclone Report, Naval Pacific Meteorology and Oceanography Centre / Joint Typhoon Center, Section 7.9, p 311. Harper B. A. and J. Callaghan, 2006: On the importance of reviewing historical tropical cyclone intensities. American Meteorological Society, 27th Conference on Hurricanes and Tropical Meteorology, 2C.1, Monterey, Apr. Harper B. A., 2002: Tropical cyclone parameter estimation in the Australian region: Wind-Pressure Relationships and Related Issues for Engineering Planning and Design, Systems Engineering Australia Pty Ltd for Woodside Energy Ltd, Perth, May, 83pp. Harper B. A., J. Kepert and J. Ginger, 2004: Guidelines for converting between various wind averaging periods in tropical cyclone conditions. WMO, TCP Sub-Project, June, 45pp, (draft). Henning, R. G., 2006: The intensity of wind gusts underneath areas of deep eyewall convection in Hurricanes Katrina and Dennis at landfall. Extended abstracts, 27th Conference on Hurricanes and Tropical Meteorology. Amer. Meteorol. Soc., Monterey, CA, April 24 - 28. Paper 3B1. Holland, G. J., 1980: An analytic model of the wind and pressure profiles in hurricanes. Mon. Wea. Rev., 108, 1212-1218. Holmes, J., K. G. Blackwell, R. A. Wade and S. K. Kimball, 2006: Collapsing precipitation cores in open-eyewall hurricanes at landfall: Are these cores actually downbursts associated with extreme surface wind gusts? Extended abstracts, 27th Conference on Hurricanes and Tropical Meteorology. Amer. Meteorol. Soc., Monterey, CA, April 24 - 28. Paper 7B7. Houston, S. H., and M. D. Powell, 2003: Surface Wind Fields for Florida Bay Hurricanes. J. Coastal Res.. 19, 503-513.

32

Kasheta, T.E. and C.B. Chang, 2002: Development of a hurricane boundary-layer wind model. Meteorology and Atmospheric Physics, 79, 259-273. Katsaros, K. B., P. W. Vachon, W. T. Liu, and P. G. Black, 2002: Microwave remote sensing of tropical cyclones from space. J. Oceanogr., 58, 137-151. Kepert, J. D., 2001: The dynamics of boundary layer jets within the tropical cyclone core. Part I: Linear theory. J. Atmos. Sci., 58, 2469-2484. Kepert, J. D. and Y. Wang, 2001: The dynamics of boundary layer jets within the tropical cyclone core. Part II: Nonlinear enhancement. J. Atmos. Sci., 58, 2485-2501. Kepert, J. D., 2002a: The impact of landfall on tropical cyclone boundary layer winds. Extended abstracts, 25th Conference on Hurricanes and Tropical Meteorology, Amer. Meteor. Soc., San Diego, California, 29 April - 3 May, 2002, 335-336. Kepert, J. D., 2002b: Modelling the tropical cyclone boundary layer wind-field at landfall. Extended abstracts, 14th BMRC Modelling Workshop: Modelling and Predicting Extreme Events, Melbourne, Australia, 11-13 November, 81-84. Kepert, J. D., 2002c: The Wind-Field Structure of the Tropical Cyclone Boundary-Layer. PhD thesis, Monash University, Melbourne, Australia. 350 pp. Kepert, J. D., 2006a: Observed boundary-layer wind structure and balance in the hurricane core. Part I: Hurricane Georges. In press, J. Atmos. Sci. Kepert, J. D., 2006b: Observed boundary-layer wind structure and balance in the hurricane core. Part II: Hurricane Mitch. In press J. Atmos. Sci. Knaff, J. A., M. DeMaria and C. R. Sampson, 2004: An introduction to the statistical typhoon intensity prediction scheme (STIPS). Extended abstracts, 26th Conference on Hurricanes and Tropical Meteorology. Amer. Meteorol. Soc., Miami, FL, May 3 - 7. Paper P14.A1. Knaff, J. A. and M. Zehr, 2006: Reexamination of tropical cyclone wind-pressure relationships. Weather and Forecasting, Apr, submitted [see also AMS 27th Hurricanes Conf]. Knupp, K. R., J. Walters, and E. W. McCaul, Jr., 2000: Profiler observations of Hurricane Georges during landfall. Geo. Res. Letters, 27, 3361-3364. Knupp, K. R., J. Walters, and M. Biggerstaff, 2006: Doppler profiler and radar observations of boundary layer variability during the landfall of Tropical Storm Gabrielle. J. Atmos. Sci. 63, 234-251. Kossin J. P., J. A. Knaff, H. I. Berger ,D. C. Herndon, T. A. Cram, C. S. Velden, R. J. Murnane and J. D. Hawkins, 2006: Estimating hurricane wind structure in the absence of aircraft reconnaissance. Weather and Forecasting, submitted Dec. Kusunoki, K., 2006: Assessment of the Doppler Radar for Airport Weather (DRAW) system in Japan as a research tool for studying typhoon. Preprints, 27th Conf. on Hurricanes and Tropical Meteorology, Amer. Meteor. Soc., P5.10 Kusunoki, K. and W. Mashiko., 2006: Doppler radar investigations of the inner core of Typhoon Songda (2004) - polygonal / elliptical eyewalls, eye contraction, and small-scale spiral bands. Preprints, 27th Conf. on Hurricanes and Tropical Meteorology, Amer. Meteor. Soc., P4.10

33

Larson, J., Y. Zhou, and R. W. Higgins, 2005: Characteristics of Landfalling Tropical Cyclones in the United States and Mexico: Climatology and Interannual Variability. J. Clim. 18, 1247-1262. Li, M., L. Zhong, W. C. Boicourt, S. Zhang, and D. -L. Zhang, 2006: Hurricane-induced storm surges, currents and destratification in a semi-enclosed bay . Geophys. Res. Lett.. 33. Lin, Y.-L., S.-Y. Chen, C. M. Hill, and C.-Y. Huang, 2005: Control parameters for the influence of a mesoscale mountain range on cyclone track continuity and deflection. J. Atmos. Phys., 62, 1849-1866. Lin, Y. -L., L. C. Savage III and C. M. Hall, 2006: Effects of landfall location and the approach angle of a cyclone encountering a mesoscale mountain range. Extended abstracts, 27th Conference on Hurricanes and Tropical Meteorology. Amer. Meteorol. Soc., Monterey, CA, April 24 - 28. Paper P4.22. Lorsolo, S. and J. L. Schroeder, 2006: Tower and Doppler radar observations from the boundary layer of Hurricanes Isabel (2003) and Frances (2004). Extended abstracts, 27th Conference on Hurricanes and Tropical Meteorology. Amer. Meteorol. Soc., Monterey, CA, April 24 - 28. Paper 10C5. Morrison, I., S. Businger, F. Marks, P. Dodge and J. A. Businger., 2005: An Observational case for the prevalence of roll vortices in the hurricane boundary layer. J. Atmos. Sci., 62, 2662-2673. Mueller, K. J., C. Miller, K. Beatty and A. Boissonnade, 2006: Correlation of topographic speed-up factors and building damage ratios for Hurricane Fabian in Burmuda. Extended abstracts, 27th Conference on Hurricanes and Tropical Meteorology. Amer. Meteorol. Soc., Monterey, CA, April 24 - 28. Paper 5A8. Nolan, D. S., 2005: Instabilities in hurricane-like boundary layers. Dyn. Atmos. Oceans, 40, 209-236. Otvos, Ervin G. 2004: Beach Aggradation Following Hurricane Landfall: Impact Comparisons from Two Contrasting Hurricanes, Northern Gulf of Mexico. J. Coastal Res. 20, 326-339. Powell M. D, P. J. Vickery and T. A. Teinhold, 2003: Reduced drag coefficient for high wind speeds in tropical cyclones. Nature, 422, 279-283. Powell, M., D. Bowman, D. Gilhousen, S. Murillo, N. Carrasco and R. St. Fleur. 2004: Tropical Cyclone Winds at Landfall: The ASOS-C-MAN Wind Exposure Documentation Project. Bull. Amer. Meteor. Soc., 85, 845-851. Rodell, M, P. R. Houser, A. A. Berg, and J. S. Famiglietti, 2005: Evaluation of 10 methods for initializing a land surface model. J. Hydrometeor., 6, 146-155. Roy Bhowmik, S. K., S. D. Kotal, and S. R. Kalsi, 2005: An empirical model for predicting the decay of tropical cyclone wind speed after landfall over the Indian region. J. Appl. Meteor., 44, 179-185. Schneider, R. and Barnes, G. M., 2005: Low-Level Kinematic, Thermodynamic, and Reflectivity Fields Associated with Hurricane Bonnie (1998) at Landfall. Mon. Wea. Rev. 133, 3243-3259. Schroeder J. L., and D. A. Smith, 2003: Hurricane Bonnie wind flow characteristics as determined from WEMITE. J. Wind Engineering and Industrial Aerodynamics, 91, 767-789. Shen, W., I. Ginis, and R. E. Tuleya, 2002: A numerical investigation of land surface water on landfalling hurricanes. J. Atmos. Phys., 59, 789-802. Shum, W. C. T. and J. C. L. Chan, 2006: Tropical cyclone landfall under the influence of uniform flow.

34

Extended abstracts, 27th Conference on Hurricanes and Tropical Meteorology. Amer. Meteorol. Soc., Monterey, CA, April 24 - 28. Paper 7B3. Shun, C. M., S. Y. Lau, and O. S. M. Lee. 2003: Terminal Doppler Weather Radar Observation of Atmospheric Flow over Complex Terrain during Tropical Cyclone Passages. J. Appl. Met.. 42, 1697-1710. Skwira, G. D., J. L. Schroeder and R. E. Peterson. 2005: Surface Observations of Landfalling Hurricane Rainbands. Mon. Wea. Rev. 133, 454-465. Stern D. P., and S. D. Aberson, 2006. Extreme vertical winds measured by dropwindsondes in hurricanes. Extended abstracts, 27th Conference on Hurricanes and Tropical Meteorology. Amer. Meteorol. Soc., Monterey, CA, April 24 - 28. Paper 16B8. Thompson E. F. and V. J. Cardone, 1996: Practical modelling of hurricane surface wind fields. J. Waterways, Port, Coastal and Ocean Engineering, ASCE, 122, 4, 195-205. Velden C., B. Harper, F. Wells, J. L. Beven, R. Zehr, T. Olander, M. Mayfield, C. Guard, M. Lander, R. Edson, L. Avila, A. Burton, M. Turk, A. Kikuchi, A. Christian, P. Caroff, P. McCrone, 2006: The Dvorak tropical cyclone intensity estimation technique: a satellite-based method that has endured for over 30 years. Bulletin American Meteorological Society, to appear. Vickery, P. J., 2005: Simple empirical models for estimating the increase in the central pressure of tropical cyclones after landfall along the coastline of the United States. J. Appl. Meteor., 44, 1807-1826. Wakimoto, R. M., and P. G. Black, 1994: Damage survey of Hurricane Andrew and its relationship to the eyewall. Bull. Amer. Meteor. Soc., 75, 189-200. Wang Y. and C. –C. Wu, 2004: Current understanding of tropical cyclone structure and intensity changes – a review. Meteor. and Atmos. Physics, 87, 257-278. Weber H. C., 2006: On the wind-pressure relationship in tropical cyclones. AMS, 27th Conf. Hurricanes and Tropical Meteorology, 14A.6, Monterey, Apr. Willoughby H. E., E. N. Rappaport, F. D. Marks, 2005a: Hurricane Forecasting, the state of the art, paper submitted for the Hurricane Forecast socioeconomic working group, Pomona CA, 16-18 February 2005. Willoughby H. E., R. W. R. Darling, and M. E. Rahn, 2005b: Parametric representations of the primary hurricane vortex. Part II: A new family of sectionally continuous profiles. Mon. Wea. Rev., 134, (4), 1102-1120. WMO 1993: Global guide to tropical cyclone forecasting. Tropical Cyclone Programme Report No. TCP-31, World Meteorological Organization, WMO/TD No. 560, Geneva. [http://www.bom.gov.au/bmrc/pubs/tcguide/globa_guide_intro.htm] WMO, 1998: Topic chairman and rapporteur reports of the fourth WMO international workshop on tropical cyclones (IWTC-IV). World Meteorological Organisation, WMO/TD - No. 875., April. Wong, M. L. M. and J. C. L. Chan, 2006a: Effects of land-sea roughness contrast on tropical cyclone winds. Extended abstracts, 27th Conference on Hurricanes and Tropical Meteorology. Amer. Meteorol. Soc., Monterey, CA, April 24 - 28. Paper 11B6. Wong, M. L. M. and J. C. L. Chan, 2006b: Tropical cyclone motion in response to land surface friction. J.

35

Atmos. Sci., 63, 1324 - 1337. Wroe, Derek R., Barnes, Gary M. 2003: Inflow Layer Energetics of Hurricane Bonnie (1998) near Landfall. Mon. Wea. Rev. 131, 1600-1612. Wu, C. -C., K. -H. Chou, H. -J. Cheng, and Y. Wang, 2003: Eyewell contraction, breakdown and reformation in a landfalling typhoon. Geophys. Res. Lett., 30, 1887. Wu, C. -C. and H. -J. Cheng, 2006: The dynamics of the eyewall evolution in a landfalling typhoon. Extended abstracts, 27th Conference on Hurricanes and Tropical Meteorology. Amer. Meteorol. Soc., Monterey, CA, April 24 - 28. Paper 6B4. Wu, M. C., W. L. Chang and W. M. Leung. 2004: Impacts of El Niño-Southern Oscillation Events on Tropical Cyclone Landfalling Activity in the Western North Pacific. J. Clim. 17, 1419-1428. Wurman, J., C. Alexander, P. Robinson and F. Masters, 2006: Preliminary comparison of DOW and in situ wind measurements in Hurricane Rita. Extended abstracts, 27th Conference on Hurricanes and Tropical Meteorology. Amer. Meteorol. Soc., Monterey, CA, April 24 - 28. Paper 10C6. Wurman, J., and J. Winslow., 1998: Intense sub-kilometer-scale boundary layer rolls observed in Hurricane Fran. Science, 280, 555-557.

36

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES Topic 0.3 : Observations and Forecasts of Rainfall Distribution Rapporteur: Lianshou Chen Chinese Academy of Meteorological Sciences 46 Zhongguancun South Street Beijing, 10081, China E-mail: lschen@cams.cma.gov.cn Fax: 86-10-62175931 Working Group: Akihiko Murata, Yihong Duan, Duong Lien chau, Ying li, Peter Black, Minghu Cheng Abstract Tropical cyclone rainfall rate and distribution is a most important issue for tropical cyclone landfall area. High impact weather and extreme events will be closely connected with rainfall brought about by landfalling tropical cyclones. There are several parts of rainfall observation forecasting techniques and physical processes etc. to be included in this report. Remotely sensed observation is a most important means for quantitative precipitation estimation (QPE) of the tropical cyclone rainfall. It includes TRMM.SSM/I, AMSU and infrared and visible data from satellite as well as the doppler radar reflectivities. On the other hand, conventional surface observation, dense spread rain gauges and automated weather stations can provide tropical cyclone rainfall basic data. To mix together those data from different sources with certain data processing techniques could provide quantitative precipitation estimation (QPE) with a certain error. A variety of forecasting techniques have been employed in quantitative precipitation forecasting (QPF) associated with landfalling tropical cyclones. Limited area models or meso scale models have been developed and used in operational forecast by many tropical cyclone forecasting centers around the world. Remotely sensed data assimilation has improved the initialization of the model and the rainfall forecasting accuracy for some of the forecasting centers. Furthermore, ensemble forecast method has been designed and employed for rainfall forecast in some of the centers as well. The effectiveness and improvements of ensemble forecasting would be evaluated in near future. Statistical scheme and statistical combined with dynamical approaches have also been used in tropical cyclone rainfall forecasting which could provide certain basic background to QPF. Empirical model is quite valuable for the forecast of rainfall distribution and intensity of landfalling tropical cyclones. Some rainfall mechanisms or physical concepts are being considered in model design. Some studies on tropical cyclone rainfall have been carried out in different institutes. Some valuable mechanisms related to rainfall distribution and intensity are found from those researches which are related to boundary layer transfer, surface and topography effect, meso scale system genesis and growth, energy fluxes and budget, extratropical transition (ET) processes, interaction between different motion scale or between different latitude systems etc. This would back up the improvements of tropical cyclone rainfall forecasting. 0.3.1 Introduction Tropical cyclone landfall involves several important issues including structure / intensity change and track turning when it approaches coastal area, sustention and decay over land, storm surge, wind strength and rainfall. Landfalling tropical cyclone rainfall is one of the most complex issues. Interaction

37

among ocean, atmosphere and land topography should be considered when tropical cyclone approaches the coastal area. Most of the severe disasters or catastrophes were often caused by heavy rainfall of a landfalling tropical cyclone which could result in flash flooding, reservoir collapses and debris flow to threaten the loss of lives and properties. Especially in recent two years, extreme weather events and disasters from landfalling tropical cyclone occurred in both Pacific and Atlantic coastal regions frequently, such as super typhoons Rananim (2004), Haitang (2005), Matsa (2005) and Saomai (2006); super hurricanes Ivan (2004), Jeanne (2004) and Katrina (2005), Rita (2005) etc. Some of the damage was caused by heavy rainfall from those typhoons and hurricanes. In this connection, accurate forecasts and warnings on tropical cyclone rainfall will play an important role for disaster prevention and preparedness. Tropical cyclone rainfall forecasting techniques are much more lagging behind to the track forecast. But significant progress has been made in recent years because the remotely sensed techniques and numerical model as well as the data assimilation techniques have been developed speedily. Several years ago, forecasts on rainfall distribution and intensity associated with tropical cyclone landfall depended on the forecaster’s subjective estimation along with their experiences. But now, the quantitative precipitation forecast (QPF) of tropical cyclone has been developed in many forecasting centers around the world. 0.3.2 Tropical Cyclone Rainfall Observations Remotely sensed data from satellite and radar are quite effective to reflect the rainfall of intensity and distribution. The Tropical Rainfall Measuring Mission (TRMM) is the first meteorological satellite loading Precipitation Radar (PR). It together with on board Visible and Infrared Scanner (VIRS) and TRMM Microwave Imager (TMI) provide a powerful capability to observe cloud precipitation. 19 landfalling tropical cyclones were selected to study the rainfall observation (Cheng 2006) with TRMM PR data measuring average area percentage (A.P) and the precipitation intensity (API) for different type of precipitation. The results show that the precipitation from both convective and stratiform are nearly the same probability density functions before and after tropical cyclone landfall. Their case studies also show that the precipitations derived from digital reflectivity of TRMM were agreeable well with that from ground radar observations. The tropical cyclone quantitative precipitation estimation (QPE) has been developed and put into operational use with Fy-2C satellite digital products by NSMC1) China since 2005. The relationship between tropical cyclone precipitation measured by surface observation and digital data from Fy-2C was set up. TBB data and cloud classification can be utilized to analyze the development and distribution of convective cloud bands and rainfall distribution. SSM/I and AMSU data are also useful to estimate the potential maximum rainfall of the landfalling tropical cyclones. They set up the cloud profile data base with microwave data. Surface precipitation can be calculated from a cloud profile selected from the profile data base which is analogous to the observed one. Radar reflectivities have widely been used to estimate the rainfall intensity and distribution of tropical cyclones. Some techniques have been developed to find the relationship between radar rainfall and true rainfall. The rainfall could be calculated from certain algorithms from relationship between radar reflectivities and observed rainfall. But the algorithms vary with different locations and rainfall properties. Some methodologies blending rain gauge data and radar data have been developed to estimate the rainfall rate and distribution (C. Velasco-Forero 2006) 1) National Satellite Meteorological Center

38

However, a lot of uncertainties still exist in precipitation estimation with either satellite data or radar reflectivities. The sources of these errors are various from both systematic and random errors. Those road blocks should be overcome through further development. 0.3.3 Tropical Cyclone Rainfall Forecast Limited area model or meso scale model with advanced data assimilation has become a major methodology for tropical cyclone quantitative precipitation forecast (QPF). A non-hydrostatic meso scale model (MSM) is being run in Japan Meteorological Agency (JMA) to predict several phenomena such as heavy rainfall. The result is used by Very Short Range Forecast (VSRF) of precipitation which provides 6-hour quantitative precipitation forecast updated every 30 minutes (Hara 2006). The rainfall forecast is based on extrapolation of latest observed precipitation in first 3 hours then the model results is combined with extrapolation in the later half. The weight of MSM prediction is increased in the forecast time afterwards since the extrapolation capability is rapidly decreased. JMA has upgraded MSM in March 2006. The horizontal resolution was increased from 10km to 5km. Daily runs were increased from 4 times to 8 times. Another QPF scheme was developed by Shanghai Typhoon Institute. They use the satellite IR/TBB data blending with hourly rain gauge data to found the QPE method for landfalling tropical cyclones. Based on QPE, very short term of 3 hour tropical cyclone rainfall forecast can be implemented with extrapolation. Rainfall ensemble forecasting systems have been designed and developed in many meteorological centers around the world. It could provide the probability of the occurrence on strong rainfall amounts. A super ensemble prediction system based on non-hydrodynamic meso scale model (MM5) has been developed by Institute of Plateau Meteorology China for rainfall forecast. The multi-physical perturbation scheme and multi-initial condition perturbation method were adopted to focus on the uncertainty of heavy rainfall event forecast in East Asia. Statistical-dynamical approaches are also employed in some of the forecasting centers which could provide rainfall forecast as well. NCHMF1) Vietnam has developed a model output dynamic (MOD) approach which utilizes dataset of the output from numerical prediction model and rain observational data. MOD provides 6 hourly rain amount and its distribution forecast. Statistical methods could provide a longer period (1-2 days) rainfall trend forecast or accumulated amount precipitation forecast with some techniques to select some analogue cases from data base. Certain criteria such as season, track and landfalling spot, topography characteristics etc. need to be set to select the analogue samples. The rainfall accumulated amount and its geographical distribution and other information from the analogue cases with certain data processing techniques could provide tropical cyclone rainfall forecast. Statistical methods also can provide a climatic background for rainfall forecast, such background can be derived from the historical data statistics. The basic background from statistics is valuable to the tropical cyclone rainfall forecast. The real time operational forecasting on rainfall rate and distribution in many forecasting centers would have a comprehensive manner. Based on tropical cyclone accurate intensity and track forecast, rainfall climatic background should be checked over. Numerous numerical models such as global model, limited area model, ensemble model etc. produce the products related to tropical cyclone rainfall prediction which should be put into comparison and checked with the satellite remotely sensed data, radar relfectivities, densely distributed rain gauge data and other QPE products. Anyway, the forecaster’s practical experiences and empirical concept are still valuable to tropical cyclone rainfall forecast. The final decision is usually made from the result of comprehensive consideration with those 1) National Center of Hydrological and Meteorological Forecast

39

QPF and QPE products. Rainfall forecast will not be convertible to flood forecast if without hydrological model. There are several important uncertainties in the hydrological model forecast. Rainfall distribution and rate forecast and surface runoff / precipitation amount are the sensitive initial conditions for hydrological model to predict flood. It shows that flash flood forecasts are heavily dependent on tropical cyclone rainfall rate and distribution forecast. 0.3.4 Physical Processes for the Tropical Cyclone Rainfall The rainfall intensity and distribution associated with landfalling tropical cyclones are closely related to the flowing physical processes. 0.3.4.1 Moisture Supply Basic factor of the tropical cyclone rainfall is the moisture transportation which usually can be distinguished from the satellite images or radar reflectivities. Tropical cyclone heavy rainfall would be stronger with a moisture supplying channel connected than that without a moisture channel. Typhoon Meari (2004) made landfall in Kyushu Japan at 08:30 (JST) 29 Sep. Heavy rainfall was occurred in Kii Peninsula even 500km far away eastward from the center of Meari. A set of numerical simulation was performed (Murata 2006) with Japan Meteorological Agency Non-Hydrostatic Model (JMANHM) with grid spacing of 5km. Five sets of initial data were used: 2100 JST 27 Sep. and 0300,0900, 1500, 2100, JST 28 Sep. The results show that the heavy rainfall was well simulated in two experiments and weak rainfall in other three. The major difference between the two group simulations is the group with heavy rainfall reproduced have relatively large amount of precipitation water compared with another group of simulation. It suggests that moisture plays a critical role in the occurrence of the heavy rainfall in both real and model atmosphere. Huge lakes, rivers, enormous reservoirs in land surface are also the sources to transfer the moisture to landfalling tropical cyclone when it covers those water source areas. A numerical investigation (Shen W. Ginis 2002) showed that the water surface over land would be favorable to decrease the decaying rate of the landfalling hurricanes. The decaying rate of the landfall tropical cyclone over water surface on land would have direct ratio with the water depth. On the other hand, a numerical simulation (Ying li et al. 2005) also showed that the moisture flux transfer of boundary layer over saturated wet soil ground would be favorable to landfalling tropical cyclone sustaining over land. Those land surface processes transfer the latent heat energy to the remnant of landfalling tropical cyclones which are also favorable to increase the rainfall of the cyclone over land. This is a rainfall feedback effect to increase further precipitation. 0.3.4.2 Extratropical Transition (ET) Some investigations exhibit that heavy rainfall caused by a weak remnant of landfalling tropical cyclone may exceed the rainfall caused by strong typhoon in its landfalling stage. It is uncertain that the rainfall has direct ratio with tropical cyclone intensity. Many cases demonstrate that landfalling tropical cyclone would be dissipated sooner or later except it obtains baroclinic energy from the mid-latitude in an extratropical transition process. Interaction between the tropical cyclone and mid-latitude weather system is an important topic including tropical cyclone extratropical transition. Observational and numerical simulation studies (Lei, 2004, Zeng 2002) showed that a tropical depression produced 304mm/24h rainfall exceeding most of the rainfall from typhoons. This rainfall was created in the interaction between the tropical depression and cold wave from the mid-latitude or in the process of extratropical transition of the tropical depression. Other study (Ying Li 2005) showed that the heaviest rain (1062/24 hours) of the remnant of

40

typhoon Nina occurred under the interaction between the cold air and the remnant of Nina. The rainfall rate was much great than the rainfall produced by super typhoon Nina in the landfall stage. Those studies show that the remnant of a landfalling typhoon would be revival or resuscitating in its ET process. It would heavily increase rainfall. Rainfall increase due to interaction between tropical cyclone and cold wave occurs not only in mid-latitude but in lower latitude as well. It is indicated that tropical cyclone would bring about large amounts of rainfall if it is associated with cold air invasion in Viet Nam. The heaviest rainfall occurs when the time of tropical cyclone landfall coincides with cold air (northeast monsoon) invasion or this invasion occurs 12-24 hours after the tropical cyclone landfall (Duong 2006). Appropriate cold air intrusion would provide remnant vortex with baroclinic potential energy converting kinetic energy and increasing potential instability. It would be favorable to increase rainfall. On the other hand, rainfall of tropical cyclone would be suppressed if more strong cold air was intruded. It could fill up the remnant. 0.3.4.3 Topographic Effect Coastal topography and mountainous range play an important role for rainfall associated with landfalling tropical cyclone. Topography convergence strengthening ascending motion in the mountain slope against wind would be a basic contribution to rainfall increase. There are three precipitation systems associated with typhoon Meari (2004) made landfall in Kyushu Japan at 08:30 JST 29 Sep. A sensitivity experiment was conducted with the model mentioned in (0.3.4.1). In the sensitivity experiment, the model topography over Kii Peninsula was removed. The simulating result showed that one of the three precipitation systems over Kii Peninsula area was disappeared and other two still existed. It is suggested that one of the precipitation system is associated with interaction with the mountain topography (Murata 2006). Asymmetric distribution of rainfall of landfalling tropical cyclone is related to coastal topographic effect. Huge rainfall area would occur in the region against the typhoon wind and less rainfall in the lee side region. Thus, the asymmetric rainfall distribution appears. Contrary, numerical simulation (Liang et al. 2002) shows that high rainfall could be occurred in the lee side of the coastland if proper cold air is brought by peripheral winds of the tropical cyclone to increase the instability in the area. Anyway, mountainous range with certain conditions could increase the rainfall. But large roughness and friction of the land surface would consume the energy of tropical cyclone and lead to dissipation. 0.3.4.4 Meso Scale Convective System A major part of heavy rainfall would be produced by strong meso scale convective systems even micro scale systems such as tornadoes. Statistical or observational studies show that those systems are subject to occur in the front or right front quadrant of the cyclone. Serial meso scale strong convective system activities would occur under the interaction among the proper cold air, mountainous range and remnant vortex itself. Strong convergence in low layer with strong vorticity, increased potential instability and upper level divergence that would be helpful the genesis and growth of meso scale convective systems. Tropical cyclone would be intensifying if a meso scale vortex merges with it (Chen, Luo 2004). Heavy rainfall would be increased in the merging process. 0.3.5 Summary Remotely sensed data from TRMM, VIRS, AMSU, SSMI Doppler radar reflectivity blending rain gauge data are quite efficient for the quantitative precipitation estimation of landfalling tropical cyclones. Uncertainties still exist in the rainfall amount from remotely sensed observation. Technical improvements need to be implemented for the future work.

41

Quantitive precipitation forecast of landfalling tropical cyclones has been developed. Limited area models or meso scale models are the major efficient means to predict very short range precipitation in 6 hours combined with extrapolation. Ensemble rainfall forecasting techniques are developing to improve the model forecast, empirical models and statistical schemes or statistical-dynamical combined approaches are also used in operational forecast and tropical cyclone rainfall climatological background can provide a valuable reference to the operational forecast. Physical processes of tropical cyclone rainfall have been studied by the research community. Moisture transportation, boundary layer fluxes transfer, extratropical transition, topographic effects, genesis and growth of meso scale convective systems are playing an important role to affect the rainfall caused by landfalling tropical cyclones. 0.3.6 Recommendations To improve the rainfall quantitative estimation and forecast associated with the tropical cyclone landfall, the following recommendations need to be addressed. (a) Field scientific experiments on landfalling tropical cyclones should be implemented. The scientific

objectives of those experiments could focus on rainfall, high winds, structure and structure change of a landfalling tropical cyclone including its boundary layer variation. Various satellite observations, doppler radar, wind profiler, densely deployed rain gauge, aircraft observation and other advanced devices could be employed in the program. The field experiments should be continued for several years with international technical cooperation. Field experiments could acquire sufficient data from various sources for theoretical and applied study to get better understanding on tropical cyclone rainfall mechanism and developing forecasting techniques.

(b) A demonstration program should be developed on landfalling tropical cyclone forecast and warning including rainfall forecast. The program attempt to obtain advanced techniques for tropical cyclone quantitative precipitation estimation and forecasting and complete a valid warning system which could be extended to other forecasting centers concerned.

(c) A numerical model comparison project should be conducted with international cooperation. The project could improve the meso scale model and ensemble forecasting techniques as well as the tropical cyclone rainfall QFE and QPF.

(d) A roving seminar on tropical cyclone rainfall estimation and forecasting should be encouraged.

This sort of seminar would efficiently help those relevant forecasting centers to develop and improve current rainfall forecasting techniques.

(e) Special workshops or symposia need to be organized on tropical cyclone QPE and QPF with WMO

funding. (f) In recent years, a variety of remotely sensed data from satellites and radars are increased rapidly.

Those data should be shared with or made available as widely as possible to most of the forecasting centers around the regions impacted by tropical cyclones. Those are the basic information to develop the QPE and QPF of tropical cyclone rainfall.

42

Bibliography Akihiko Murata, 2006: A cloud-resolving numerical simulation for characteristic rainfall induced by typhoon Meari (2004), Team member report on topic 0.3 of Sixth International Workshop on Tropical Cyclones C. Velasco-Forero, et al., 2006: An automatic methodology for estimation of rainfall fields blending radar and rain gauges data in real time, 2nd International Conference on Quantitative Precipitation Forecasting and Hydrology Proceedings, P. 26 Chen Lianshou, Lou Zhexian, 2004: Interaction of typhoon and meso scale vortex, Advances in Atmospheric Sciences, Vol 21. No4, 515-528 Cheng Minghu, et al., 2006: The studies of landfalling tropical cyclone using TRMM and HK radar data, Team member report on topic 0.3 of IWTC-VI Duong Lien Chau, 2006: Team member report on topic 0.3 of Sixth International Workshop on Tropical Cyclones Feng Hanzhong, et al., 2006: Verification of precipitation prediction from meso scale ensemble system for the upper basin of Yangtze River, 2nd International Conference on Quantitative Precipitation Forecasting and Hydrology. Proceedings, P.68 Kwan-Yin Kong, 2002: Anomalous intensification of the remnants of tropical storm Allison over land, 25th Conference on Hurricane and Tropical Meteorology. San Diego CA, American Meteorological Society Lei Xiaotu, 2004: A case study on the heavy rain in Shanghai induced by a tropical depression on 5 Aug. 2001, Proceedings on WMO International Workshop on Tropical Cyclone Landfall Processes, 157-159 Liang Xudong, Duan Yihong, Johnny C. L. Chan, 2002: Convective Asymmetries associated with tropical cyclone landfall, ACTA Meterologica Sinica, Vol. 60 (Supplement), 26-35 (in Chinese) Li Ying, et al., 2005: Numerical study on impacts of boundary layer fluxes over wetland on sustention and rainfall of landfalling tropical cyclone, Acta Meteorologica Sinica, 63(5), 683-693 Shanghai Typhoon Institute, 2006: Team member report on topic 0.3 of IWTC-VI Shen W. Ginis, 2002: A numerical investigation of land surface water on landfalling hurricane, J Atmos Sci., 789-802 Tabito Hara, 2006: The operational non-hydrostatic meso scale model for quantitative precipitation forecasts at JMA, 2nd International Conference on Quantitative Precipitation Forecasting and Hydrology, Proceedings, P.36 Zeng Zhihua, Chen Bomin, Gao Quanping, 2002: Simulation of influence of “urban heat island” on torrential rain in Shanghai on 5 Aug. 2001, ACTA Meteorologica CINICA, Vol 60 (Supplement) 58-64 (in Chinese).

43

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES Topic 0.4 : Observations and forecasts of storm tides Rapporteur: Shishir K. Dube, Indian Institute of Technology, Kharagpur, INDIA Email : skdube@iitkgp.ernet.in Working Group: Bruce Harper, Gabriele Gönnert, Masakazu HIGAKI, Nadao Kohno 0.4.1 Introduction The damage from land falling cyclones is mainly due to three factors: rain, strong winds, and storm tide. Storm tides associated with severe tropical cyclones are by far the most damaging. The storm tide is the combined water level associated with the simultaneous effects of the astronomical tide, storm surge and breaking wave set-up (refer Figure 0.4.1). Death and destruction arise directly from the intense winds that are characteristic of tropical cyclones blowing over a large surface of water, bounded by a shallow basin. These winds cause the sea water to pile up on the coast and leads to sudden inundation and flooding of coastal regions. About 90% of the damage is due to inundation of land by seawater. In addition, flooding of the river deltas occur from the combined effects of tides and surges from the sea penetrating into the rivers, because at the same time excess water in the rivers due to heavy rains from the cyclone is trying to flow through the rivers into the sea. On the other hand, small offshore islands in deepwater environments can be most affected by the breaking wave setup component (e.g. on fringing reefs such as TC Zoe at Tikopia in 2003 and TC Meena at Rarotonga in 2005) or wave run up (on unprotected steep deepwater coasts, e.g. the devastation caused by overtopping at Niue Island during TC Heta in 2004). Most of the world's greatest human disasters associated with the tropical cyclones have been directly attributed to storm surges. Thus, the real-time monitoring and warning of storm tide is of great concern. It is necessary that the problem of the storm tide be seriously addressed by the countries of the various regions through collective efforts and in an integrated way. It may be noted that prediction of storm surge amplitude, peak wave heights and extent of coastal inundation depend critically on the prediction of track (landfall), intensity and the spatial structure (wind, pressure) of tropical cyclones. Progress has been made in cyclone forecasting and warning during the last two decades under regional projects on tropical cyclones. The same cannot be said about the procedure for storm tide forecasting and warning. It is still inadequate. Cyclone forecasting must be further improved, and applied to drive models for forecasting storm tides. There are several aspects of the cyclone and storm tide warning systems and the disaster mitigation that require considerable improvements so as to bring about better response and minimize the loss of life and property.

44

Fig. 0.4.1: Components of a storm tide 0.4.2 Progress in Storm Tide Modelling and Forecasting There has been little new formally published material on the subject of the modelling and forecasting of storm tide since the IWTC-V in 2002. The reason for this is possibly that: a) the hydrodynamics of storm tide generation and propagation is relatively well established; b) accuracy of predictions is largely limited by the meteorological inputs, and c) implementation of new regional models is limited by lack of resources and data. The latter factor is critical in that many developing countries, typically with the greatest exposure to storm tide, lack the resources, data and funding to develop sophisticated warning systems, even though the science to do so is well established. In Australia, an upsurge in natural hazards risk assessment spending driven by the desire to understand enhanced Greenhouse threats has lead to increased development of models and systems (e.g. HARPER 2001, 2005). These developments have additionally leaded to improvements in operational storm tide warning systems. JENSEN ET.AL (2005, 2006) simulated extreme storm surge events in the German Bight and computed the probabilities of their occurrence. Bibliography at the end of the report attempts to partially list the publications on storm surge since ITWC-V in 2002.

Ocean WavesOcean Waves

Wave SetupWave Runup SWLMWL

HAT

AHD

ExtremeExtremeWindsWinds

ExpectedHigh Tide

CurrentsCurrentsSurgeStormStorm

TideTide

after Harper (2001)

Ocean WavesOcean Waves

Wave SetupWave Runup SWLMWL

HAT

AHD

ExtremeExtremeWindsWinds

ExpectedHigh Tide

CurrentsCurrentsSurgeStormStorm

TideTide

Ocean WavesOcean Waves

Wave SetupWave Runup SWLMWL

HAT

AHD

ExtremeExtremeWindsWinds

ExpectedHigh Tide

CurrentsCurrentsSurgeStormStorm

TideTide

after Harper (2001)

45

0.4.3 Operational Storm Surge Prediction Models Operational numerical storm surge prediction models have been developed and are being routinely used for several coastal regions of the world, such as North Sea, the Gulf of Mexico and Atlantic coasts, Hong Kong, China, etc. Advent of powerful personal computers has set up a trend to run storm surge models in real-time on PC-based workstations in an operational office. In fact, a PC-based work station (the Automated Tropical Cyclone Forecasting System, ATCF) has already been operation at the Joint Typhoon Warning Centre, Hawaii for years. The Australian Bureau of Meteorology Research Centre, together with their Bureau Severe Weather Programme Office, has also developed an Australian workstation for storm surge forecasting. This has been augmented recently by the adoption of rapid assessment parametric storm tide models in Queensland (e.g. derived from HARPER,2001) and by a probabilistic forecast model for the Northern Australian coastline (SEA 2005ab). In India, DUBE ET AL (1994) developed a real-time storm surge prediction system for the coastal regions of India. Real-time storm surge prediction systems have also been developed for Bangladesh, Myanmar, Oman, Pakistan, Sri Lanka and Thailand. The National Meteorological and National Hydrological Services of many countries have achieved some success in provision of storm surge warnings and for implementing improved models through co–operative and co–ordinated sharing of responsibilities within the framework and overall guidance and supervision of the Tropical Cyclone Programme (TCP) of the World Meteorological Organization (WMO). The TCP of WMO supported technology transfer to Bangladesh, Myanmar, Oman, Pakistan, Sri Lanka and Thailand from the Indian Institute of Technology-Delhi/Kharagpur. However, there are still several nations that still do not have expertise and resources to run an operational storm tide model. General forecasting authority of the Federal Government of Germany attached to the Federal Maritime and Hydrographic Agency forecasts the daily tide level and storm surge. In case of storm surge they warn for the whole North Sea coast and give a prognosis for Hamburg. 0.4.4 Probabilistic storm tide forecasting

(Ensemble methods in storm tide modeling)

While preparing meteorological fields for storm tide forecasting, it is important to consider the error of tropical cyclone track forecast and its influence on storm tide forecasting. Although the performance of tropical cyclone forecast has been advancing steadily, the mean position error in track forecast today is still around 100km for 24-hour forecast and 300km for 72-hour forecast as shown in Figure 0.4.2 (JMA, 2005). This implies that there is probably a large spread of possible forecast values of surface winds and atmospheric pressure at a certain location and the spread, which makes accurate storm surge prediction difficult even for 24-hour forecast.

Figure 0.4.3 demonstrates how the difference in the path of a typhoon changes storm surge occurrence. If a typhoon takes a path left of the forecast track, a storm surge may occur in the western bay in the area shown in the figure (Figure 0.4.3(b)), while a surge may occur in the eastern bay in the figure if the typhoon takes a right path (Figure 0.4.3(c)).

To take into account the influence of tropical cyclone track on the occurrence of storm surge, some National Meteorological Services utilize ensemble or probabilistic methods in storm tide forecast.

46

Fig. 0.4.2: Annual mean position errors of 24-, 48- and 72-hour operational typhoon track forecasts

(From JMA, 2005)

(a) (b) (c)

Fig. 0.4.3: Maximum surge envelopes simulated with different typhoon tracks. (unit: cm) (a) Typhoon track used in the simulations. (b) The case in which a typhoon takes the westernmost path. (c) The same as (b) but for the easternmost path.

While probabilistic forecasting of track has become more widespread in many agencies through the use of ensemble NWP tracks, the extension to wind probabilities and especially storm tide forecasting is yet to be widely adopted. While trivial in concept, (i.e. multiple storm surge, wave and/or other associated modelling for a forecast event) the practical implementation in a forecasting environment can be difficult and the computational burden can be significant. Nevertheless, given that the accuracy of storm surge and wave modelling is largely dependent upon the accuracy of the meteorological inputs, it is clear that stochastic simulation of storm tide is an important component for an effective warning system. Besides the dependency of peak storm surge on landfall position and local bathymetry, a macro tidal environment is especially sensitive to the relative timing of the tide and the landfall. Changes in track, speed and intensity can therefore all significantly change the forecast impact. In the past this variability has been accommodated by the use of databases of precomputed MEOWs

47

(Maximum Envelope of Waters) and also the Maximums of the Maximums (MOM)s technique (e.g. FEMA 1999). While the MEOW/MOMS approach is typically conservative it does not necessarily provide the best information to emergency managers and it is difficult to apply directly in a macro tidal environment. The Darwin Tropical Cyclone Warning Centre (TCWC) has recently adopted the SEAtide probabilistic model (SEA 2005ab), which uses a parametric storm tide subsystem to enable a rapid Monte Carlo simulation of a user-specified range of storm parameters. The system includes astronomical tide and waves, with allowance for some non-linear interactions. Figure 0.4.4 shows an example of the SEAtide model applied to Townsville in North Queensland, where the 10% non-exceedance storm tide is demonstrated, the prediction being based on a sample of 20 possible forecast scenarios (100 or more scenarios can be rapidly simulated). These concepts can be readily extended to full numerical solutions if the computational power is available, or the parametric forecasts can be further verified by a deterministic storm surge.

Fig. 0.4.4: An example of the SEAtide model probabilistic storm tide output Importantly, the system provides the emergency manager with a comprehensive assessment of all threatened localities and the timing and relative impact of possible storm tide flooding. The advantage of probabilistic storm tide forecasting would be that the full range of viable possibilities is explored (track, speed, intensity, scale, tide, timing etc) and the forecaster can focus on which specific meteorological parameters are the least well known and/or will have the greatest effect on a specific storm tide forecast in a specific area of coast. Japan Meteorological Agency (JMA) conducts five runs of the storm surge model with five possible typhoon tracks in its storm surge prediction system. These five typhoon tracks are prescribed at the center and at four points on the forecast circle within which a typhoon is forecasted to exist with a probability of 70%, and used to make meteorological fields with the empirical typhoon model. Having a sufficient number of ensemble members, the ensemble will provide probabilistic information on storm surges. For example, NOAA, U.S.A. starts the provision of “The experimental Probabilistic Hurricane Storm Surge product” on their web page since the beginning of this hurricane season. The product consists of two graphics for the Gulf of Mexico and the Eastern Atlantic coastal areas (Figure 0.4.5). One shows probabilities, in percent, of storm surge exceeding 5 feet. The other indicates there

48

is a 10 percent chance of the displayed storm surge heights being exceeded. These storm surge graphics are based upon an ensemble of Sea, Lake, and Overland Surge from Hurricanes (SLOSH) model runs using the National Hurricane Center (NHC) official advisory and accounts for track, size, and intensity errors based on historical errors. This kind of products is expected to provide users with information, which enhances their ability to make preparedness decisions specific to their own situations. (a) (b)

Fig. 0.4.5: Examples of “The experimental Probabilistic Hurricane Storm Surge products”(From NOAA

web site, http://www.weather.gov/mdl/psurge/) (a) 10 Percent exceedance height, (b) Probability of storm surge exceeding 5 feed above normal tide

0.4.5 Improvements in specifying wind and pressure fields As Meteorological fields, particularly wind fields, have the biggest impact on the performance of storm tide modeling, it is of great importance to improve the accuracy of wind and pressure fields. Typically, numerical storm tide modelling systems incorporate simplified (analytical, parametric) descriptions of the wind and pressure fields of tropical cyclones (e.g. HOLLAND,1980). While such representations are consistent with the level of storm details currently provided by, for example, the worldwide use of the Dvorak technique, increasing knowledge of storm structure and behaviour leads to a desire to improve such models. More sophisticated models have become available in recent years (KEPERT 2001, Willoughby et al. 2005) that have the capacity to improve the representation of wind and pressure fields and these should be considered for future model developments. In all cases, calibration remains an essential tool when considering the relative merits of different wind and pressure models. In recent years, non-hydrostatic mesoscale NWP models with high horizontal resolution, say, 5-10km, such as NHM of JMA (SAITO ET AL., 2006) have been put into operation at several National Meteorological Services in order to improve short-range weather forecast. Because they are having a capability to represent the structure of a tropical cyclone, applying the meteorological fields derived from these NWP models is likely to improve the performance of tropical storm surge forecasting dramatically, especially for short-range surge forecast. It should be noted that those NWP models could provide sufficient wind fields for storm surge modeling only when given an initial condition that reflects the observed tropical cyclone structure correctly. The initialization can be accomplished by a so-called “bogusing” scheme and sophisticated data assimilation method like 4D-VAR.

49

Considering the current progress of NWP model technology, the feasibility of applying mesoscale NWP model results to storm tide modeling should be further investigated. While there may be advantages to some agencies in developing more complex coupled ocean and atmospheric models for improving storm intensity forecasting, the advantages in specifically extending this to include storm tide are likely to be less critical, especially if coastal data is of poor quality. The use of simplified analytical wind and pressure models continues to show good performance for storm tide modelling and, provided uncertainty in parameter estimation is addressed, many nations could continue to significantly improve their storm tide forecasting capabilities without resorting to the additional expense of developing complex coupled models. Importantly, complex models should not prevent the forecaster from adequately considering the uncertainty in the meteorological parameters.

One major benefit of coupled models however could be in the area of improving surface wave estimates due to a potentially better representation of wave fetch. The relative import of this would be site specific but likely to improve wave setup forecasts for offshore islands, where water level and peak spectral wave period can become critically important. A powerful yet simpler technique is to incorporate more sophisticated parametric wind models, such as that by HARDY ET AL. (2003).

0.4.6 Potential impacts of possible enhanced Greenhouse climate change While the potential for sea level rise under a changed climate will add to any predicted storm tide level in a direct way, the likely impacts on the meteorological parameters are still subject to intense debate. More subtle effects however may eventuate due to nonlinearities of interactions in some specific regions and situations. These possibilities should be explored within the mitigation-planning framework through the systematic consideration of possible climate change effects (e.g. HARPER,2001;2004). Detailed numerical modelling should then provide adequate allowance for these effects and forecast models can be readily adapted to meet this challenge. GÖNNERT,(2002) computed maximum storm surge curve for North Sea (Figure 0.4.6) due to the global warming in the 20th century with an increase of the temperature of 0.6 to 0.9° C. The numerical modelling considers the projected global warming from the Intergovernmental Panel of Climate Change (IPCC, 2001; GROSSMANN, et.al. 2007).

surge

storm tide curve

surge curve

mean tide curveastronomical or

Fig. 0.4.6: Storm Tide and Storm Surge Curve in the North Sea (GÖNNERT, 2002)

50

Recently there have been some studies (WOTH, 2005; WOTH, K. and STORCH, H. VON, 2006; WOTH ET AL., 2006) to look into the possible changes in North Sea storm-related water heights due to increasing greenhouse gases. 0.4.7 Workshops and Hands-on Forecast Training Laboratories Storm Surge and Wave Forecasting (WMO/JCOMM Initiatives) JCOMM/TCP took initiatives, under the leadership of Dr Johannes Guddal, to organize series of Workshops on storm surge and ocean waves and hands on training for the countries of South China Sea and also Indian Ocean. The overall aim of the workshop was to enable these countries, through technological and scientific progress and mutual cooperation, to establish and/or to improve their systems of marine forecasting, in particular with regards to coastal storm surges connected with tropical cyclones. Except the first workshop subsequent workshops were structured as a hands-on training laboratory for ocean forecasting (Ocean waves and storm surges) At the end of the workshops forecasting models were transferred to the participants to enable trainees to run operational wave and storm surge forecasting in their respective home countries. Workshops and hands-on training have been great success. The detailed reports of the workshops can be found out at http://www.wmo.int/web/www/TCP/TCP-home.html. List of the workshops organized is given below:

1. JCOMM/TCP Workshop on South China Sea Storm Surge, Wave and Ocean Circulation Forecasting (Hanoi, Viet Nam 21 – 24 January 2002).

2. The Second Workshop on South China Sea Storm Surges, Waves and Ocean Circulation

Forecasting “A Hands-on Ocean Forecast Training Laboratory for the South China Sea Region” Kuantan, Malaysia, 15 – 19 September 2003.

3. Third Regional Workshop on Storm Surge and Wave Forecasting – A Hands-on Forecast

Training Laboratory (Beijing, China, 25 to 29 July 2005).

4. Fourth Regional Workshop on Storm Surge and Wave Forecasting - A Hands-on Forecast Training Laboratory (Manila, 11 to 15 September 2006).

0.4.8 Recommendations

a) Storm surge forecasting and warning systems are not adequate in many nations. Particular attention is urgently needed to develop models for different coastal zones.

b) Storm surge predictions are readily affected by the error in tropical cyclone predictions in terms

both of their tracks and of intensities. Taking into account this, ensemble (-like) and probabilistic methods and outputs should be considered to use in operational storm surge forecast.

c) As mesoscale NWP models with high resolution are having ability to solve tropical cyclone

fields, the use of the results of these NWP models in tropical storm surge modeling should be investigated.

d) The most important need is the development of robust and reliable operational technique for prediction of storm surge – based on sound hydrodynamics in numerical models. Particular

51

attention needs to be given to the coastal regions taking into account the complex coastal orientation and estuaries, and this includes their massive freshwater discharge.

e) Total storm tide water level is the combined effect of storm surge, wave set-up and high tide,

and so accurate prediction of wind waves and tidal height together with their non-linear interaction with the storm surge in the model is essential, but prediction methods for wave setup are not well established yet. Therefore, further studies of wave setup prediction method are needed.

f) While accurate prediction of surge is important, it is also necessary to have an estimate of the

coastal stretch likely to be inundated for effective evacuations. To achieve this effort, to develop real-time ocean-river coupled models are required for the regions where they do not exist. For the Bay of Bengal region, this aspect is very important as one of the world’s largest river systems (Ganga-Brahmputra-Meghana) joins the head Bay of Bengal (DUBE et al. 1986; DUBE ET AL. 2005). The Southeast River Forecast Center in the USA is working with the University of Central Florida to use tide forecast and ultimately storm surge forecasts, in conjunction with the National Weather Service River Forecast System. A dynamic routing model is proposed to be applied and the river-level forecast is expected to improve under normal tides and storm surge conditions.

g) Another area that requires attention is the impact of climate change and a possible sea-level

rise and changes in the frequency and intensity of storms. These factors may change flooding risks from storm surges, especially in the low-lying regions of the world such as Bangladesh and Maldives.

h) Detailed time histories and data dossiers of individual storm surges should be prepared by the

concerned countries, which will enable calibration of storm surge models and improvement of prediction techniques. These data are also helpful to assess the potential and susceptibility of the coastlines. Estimates of storm surge potential from historic records are also valuable for efficient administration of cyclone mitigation plans to determine the safety and economics of coastal construction and installations.

i) It also would also be appropriate to store all the pertinent data when a given storm affects an

area, i.e., inundation maps, high-water marks, etc. It is also important to mention that now GIS (Geographical Information System) work is a common tool for most researchers, the design and creation of a GIS that contains precipitation, stream flow, and hurricane track data would be very valuable.

j) The response of the public and disaster preparedness agencies must be strengthened through

better scientific understanding of cyclones and storm surges, and their warnings and related information. This is the area that requires great attention. In spite of some improvement in warning systems, adequate attention has not been given in the development of mechanisms for public response and disaster preparedness organizations. Reliable forecasts lead to public confidence and positive responses to warnings. To get a better response from the public, it is necessary to disseminate this information in a language and wordings that the public can understand, so threats can be properly conveyed to them. Effort is also needed to strengthen communication systems together with warning systems in the regions.

k) Capacity building and development of human resources in all facets of the storm surge

problem is the most important area, which should be given great attention to achieve self–sufficiency by the nations, which lack expertise. Short-term orientation courses for operational forecasters, plus regular workshops and seminars should be organized in the regions.

52

0.4.9 Summary This report includes progress and improvements of the storm surge models (including inland inundation). Developments are proposed in effective operational numerical storm surge prediction facility for forecasting the total water level by including non-linear interaction of wind waves and astronomical tides. A key component of the recommendation is capacity building and human resources development in the region of the vulnerable nations through training, provision of transfer of technology, and workshops / seminars. Bibliography

BAERENS, C. (2003): Die Wasserstände an der Ostseeküste: Entwicklung, Sturmfluten, Klimawandel.

BARDOSSY, A., PAKOSCH, S. (2005): Wahrscheinlichkeiten extremer Hochwasser unter sich ändernde. Klimaverhältnissen. Wasserwirtschaft, Heft 7-8, Wiesbaden.

BÄRRING, L., H. VON STORCH (2004): Northern European Storminess since about 1800. Geophysics. Res. Letters 31. BOON, JOHN D. (2004): Secrets of the tide: tide and tidal current analysis and applications, storm surges and sea level trends. Chichester, U.K CHITTIBABU, P., S. K. DUBE, P. C. SINHA, A. D. RAO, AND T. S. MURTY, (2002). Numerical simulation of extreme sea levels for the Tamilnadu (India) and Sri Lanka coasts. Marine Geodesy 25, 235-244. CHITTIBABU, P, S. K. DUBE, J. B. MACNABB, T. S. MURTY, A. D. RAO, U. C. MOHANTY AND P. C. SINHA, (2004). Mitigation of flooding and cyclone hazard in Orissa, India. Natural Hazards 31 (2004) 455-485 DUBE, S.K., P.C. SINHA, AND G.D. ROY, (1986). Numerical simulation of storm surges in Bangladesh using a River - Bay coupled model. Coastal Engineering, 10, 85-101. DUBE, S.K., A.D. RAO, P.C. SINHA, AND P. CHITTIBABU, (1994). A real-time storm surge prediction system: An application to east coast of India. Proc. Indian Natn. Sci. Acad. 60, 157-170. DUBE S. K., P. CHITTIBABU, P. C. SINHA, A. D. RAO AND T. S. MURTY, (2004). Numerical modeling of storm surges in the head Bay of Bengal using location specific model. Natural Hazards 31 (2004), 437-453

DUBE, S. K., GÖNNERT, G., MUNROE, A., MURTY, T. S., PADALA, C., RAO, A.D., SINHA, P. C. (2004): Storm Surges from Extra-Tropical Cyclones. In: Rao, A. D., Gönnert, G. (Ed.): Storm surges, sedimentation and coastal erosion, Natural Hazards, (32) 2, Special Issue. DUBE, S. K., P. C. SINHA, A. D. RAO, INDU JAIN AND NEETU AGNIHOTRI, (2005). Effect of the Mahanadi river on the development of storm surge along the Orissa coast of India: A numerical study. Pure appl. Geophys. 162 (2005) 1673-1688. EMANUEL K. (2003): A Similarity Hypothesis for Air Area Exchange at Extreme Winds Speeds. Journal of Atmospheric Sciences, 60. FEMA (1999) Northwest Florida hurricane evacuation study. Prep by FEMA, NOAA, State of Florida and US Army Corps of Engineers, July. FISCHER-BRUNS, I., H. VON STORCH, F. GONZÁLEZ-ROUCO AND E. ZORITA. (2005): Modelling the variability of multitude storm activity on decadal to century time scales, Geesthacht.

53

GISZAS, H. (2004): Sturmflutschutz: Herausforderung und Sicherheitskonzepte. Hansa,141.Jahrgang Nr. 2, Februar 2004, Hamburg.

GÖNNERT, G. (2002): Risk assessment by calculating a maximum storm surge curve in Hamburg due to global climate change during the 20th century. In: Deutsches IHP-OHP - Nationalkomitee (Ed.): Low-Lying Coastal Areas - Hydrology and Integrated Coastal Zone Management, Koblenz, 2002, S. 73-78.

GÖNNERT, G. (2004): Sturmfluten und Windstau in der Deutschen Bucht. Charakter, Veränderungen und Maximalwerte im 20. Jahrhundert. In: Die Küste, 67.

GÖNNERT, G., (2004): Maximum storm surge curve due to the global warming during the 20th in the European North Sea Region (German Bight). In: Rao, A. D., Gönnert, G. (Ed.): Storm surges, sedimentation and coastal erosion, 2004, Special Issue. Natural Hazards, (32) 2, Special Issue S. 211-218.

GÖNNERT, G.(2004) : Coastal Protection in Hamburg. In: Schernewski, G., Dolch, T.: Geographie der Meere und Küsten. Coastline Reports des EUCC 2004-1.

GÖNNERT, G.(2004): Sturmfluten und Windstau in der Deutschen Bucht. Charakter, Veränderungen und Maximalwerte im 20. Jahrhundert. In: Die Küste, 67, 2004

GÖNNERT, G., GIESE, H., PLÜß, (2004), A.: Charakterisierung der Tidekurve. In: Die Küste, 68. GÖNNERT, G., GRAßL, H., KUNZ, H., KELLETAT, D., VON STORCH, H., SÜNDERMANN, J. (2004): Klimaänderung und Küstenschutz. Proceedings zur Tagung vom 29. und 30.11.2004, Hamburg.

GÖNNERT, G., GÜNTHER, H., BEHRENDT, M.(2005): Sturmflutsicherheit in Hamburg angesichts eingetretener sehr hoher Oberwasser vor dem Hintergrund des Augusthochwassers im Jahre 2002. In: Tagungsband der Konferenz der Hafenbautechnischen Gesellschaft 2005. Hamburg. GROSSMANN, I., K. WOTH AND H. VON STORCH. (2006): Localization of global climate change: Storm surge scenarios for Hamburg in 2030 and 2085.

HANS KUNZ. (2005): Statements zu Küstenschutz und Risikobewältigung: Sprechtag zum Thema "Klimaänderung und Küstenschutz". Hansa, Bd. 142. HARPER B.A., (2004) Guidelines for Responding to the Effects of Climate Change in Coastal and Ocean Engineering – 2004 Update. Engineers Australia, National Committee on Coastal and Ocean Engineering, EA Books, 76pp. [see also http:/www.engineersaustralia.org.au/nccoe/publicat.htm] HARPER B.A., (2005) Recent advances in storm tide modelling in Australia, WMO International Workshop on Tropical Cyclone Landfall Processes, p17-21, Macau, China, March. HAGEN, S.C., O. HORSTMAN, AND R.J. BENNETT,(2002). “ An Unstructured Mesh Generation Algorithm for Shallow Water Modeling.” The International Journal of Computational Fluid Dynamics”, 16, 83-91. HARDY, T.A., MCCONOCHIE, J.D. AND MASON, L.B., (2003) Modelling tropical cyclone wave population of the Great Barrier Reef. J. Waterway, Port, Coastal and Ocean Engineering, ASCE, 129 (3), 101-113.

HOFSTEDE, J. (2005): Comrisk: common strategies to reduce the risk of storm floods in coastal lowlands. The Coast, Bd. 70. HUPFER, P. (2003): Die Wasserstände an der Ostseeküste: Entwicklung - Sturmfluten – Klimawandel. Cuxhaven: Kuratorium für Forschung im Küsteningenieurwesen ISERT, K., GÖNNERT, G., GIESE, H. (2003): Analyse der Tidekurve. In: Daschkeit, A./ Sterr, H., 2003: Aktuelle Ergebnisse der Küstenforschung. 20. AMK Tagung Kiel. Berichte, Forschungs- und Technologiezentrum Westküste d. Univ. Kiel, 28. Kiel.

54

JAPAN METEOROLOGICAL AGENCY (2005): Annual Report on Activities of the RSMC Tokyo –Typhoon Center, http://www.jma.go.jp/jma/jma-eng/jma-center/rsmc-hp-pub-eg/AnnualReport/2004/Text/Text2004.pdf

JENSEN, J. (2003): Neue Verfahren zur Abschätzung von seltenen Sturmflutwasserständen. Hansa, Bd. 140.

JENSEN, J. (2005): Abschlussbericht: Modellgestützte Untersuchungen zu Sturmfluten mit sehr geringen Eintrittswahrscheinlichkeiten an der Deutschen Nordseeküste (MUSE). Siegen.

JENSEN, J., FRANK, T. (2003): Abschätzung von Sturmflutwasserständen mit sehr kleinen Überschreitungswahrscheinlichkeiten. In: Die Küste, Heft 67.

JENSEN, J., MUDERSBACH, CH. (2003): Untersuchungen zur Wasserstandsentwicklung an der dt. Nord- und Ostseeküste bis 2001. Bericht im Auftraf des BfG.

JENSEN, J, MUDERSBACH, CHR. (2006): Personal Communication, Hamburg

KATZ, R.W., PARLANGE, M.B. UND NAVEAU, P. (2002): Statistics of extremes in hydrology, Advances in water resources 25. Elsevier. KEPERT, J. (2001) The dynamics of boundary layer jets within the tropical cyclone core - part I: linear theory. Jnl Atmospheric Sciences, 58, Sept, 2469-2484. KRAUS E. B., EBEL U. (2003): Risiko Wetter. Die Entstehung von Stürmen und anderen atmosphärischen Gefahren. KUNZ, H. (2004): Sturmflutschutz und Küstenmanagement. HTG-Kongress 2003, Hansa 141. Jahrgang, Nr.2, Hamburg.

LASSEN, H., SIEFERT, W., GÖNNERT, G. (2002): Windstauentwicklung in dem Tiefwasserbereich der Südöstlichen Nordsee bei Sturmflutwetterlage. In: Die Küste, 64. LOEWE P. (2005): Nordseezustand 2003. Report of Federal Maritime and Hydrographic Agency, Nr. 38.

MÜLLER-NAVARRA. (2006): Sturmflutvorhersagen für die Nordsee. Sturmfluten sind durch meteorologische Einflüsse verstärkte Fluten.www.dwd.de/de/Zusatzmenues/ Presse/Mitteilungen/20060215e.htm - 8k –

MÜLLER-NAVARRA, S. H., LANGE, W., DICK, S., SOETJE, K. C. (2003): Über die Verfahren der Wasserstands-und Sturmflutvorhersage: Hydrodynamisch-numerische Modelle der Nord- und Ostsee und ein empirisch-statistisches Verfahren für die Deutsche Bucht. Promet, 29.

PLÜß, A. (2003): Das Nordseemodell der BAW zur Simulation der Tide in der Deutschen Bucht. In: die Küste, Heft 67. PUGH, DAVID T. (2004): Changing sea levels: effects of tides, weather and climate. Cambridge University Press, Printout REESE, STEFAN (2003): Die Vulnerabilität des schleswig-holsteinischen Küstenraumes durch Sturmfluten: Fallstudien von der Nord- und Ostseeküste. Forschungs- u. Technologiezentrum Westküste, Büsum. SAITO, K., T. FUJITA, Y. YAMADA, J. ISHIDA, Y. KUMAGAI, K. ARANAMI, S. OHMORI, R. NAGASAWA, S. KUMAGAI, C. MUROI, T. KATO, H. EITO AND Y. YAMAZAKI (2006): The Operational JMA Nonhydrostatic Mesoscale Model. Monthly Weather Review, 134, 1266–1298.

55

SCHWENN, VERENA (2005): Zur Gefährdung der Nord- und Ostseeküsten sowie der Schifffahrt durch große Sturmfluten. Diploma Thesis Sommersemester 2005. SEA (2005a) Darwin TCWC Northern Region Storm Tide Prediction System – System Development Technical Report, Prep by Systems Engineering Australia Pty Ltd for the Bureau of Meteorology, Darwin, 208pp, Dec. SEA (2005b) SEAtide V1.1 Users Guide. Systems Engineering Australia Pty Ltd, 46pp, Dec. SZTOBRYN, M. (2005): Sturmfluten in der südlichen Ostsee (westlicher und mittlerer Teil).Maritime and Hydrographic Agency, Hamburg. TRIEBNER, J., GÖNNERT, G. (2004): Besonderheiten des Sturmflutschutzes in urbanen Ästuargebieten am Beispiel Hamburgs. In: Gönnert, G., Graßl, H., Kunz, H., Kelletat, D., Von Storch, H., Sündermann, J. (Ed.): Klimaänderung und Küstenschutz. Proceedings zur Tagung vom 29. und 30.11.2004, Hamburg.

WEISSE, RALF (2005): Marine Climate Change: Ocean Waves, Storms and Surges in the Perspective of Climate Change. WEISSE, R. PLÜß, A. (2005): Storm related sea level variations along the North Sea Coast as simulated by a high-resolution model. Ocean Dyn. 1958-2002. WEISSE, R., VON STORCH, H. AND FESER, F. (2005): Northeast Atlantic and North Sea storminess as simulated by a regional climate model 1958-2001 and comparison with observations. J. Climate 18. Willoughby, H.E., Darling, R.W.R. and Rahn, M.E., (2005) Parametric representations of the primary hurricane vortex. Part II: A new family of sectionally continuous profiles. Mon. Wea. Rev., 134, (4), 1102-1120.

WOTH, K. (2005): Projections of North Sea storm surge extremes in a warmer climate: How important are the RCM driving GCM and the chosen scenario? Geophysics. Res. Let., vol32. WOHT, K. (2006) :Ppersonal Communication. WOTH, K., STORCH, H. VON. (2007): Personal Communication, Hamburg. WOTH, K., WEISSE, R. AND VON STORCH, H. (2006): Dynamical modelling of North Sea storm surge extremes under climate change conditions. An ensemble study. Ocean Dynamics,Vol56.

56

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES Topic 0.5 : Observations And Forecasts Of Hydrology-Related Disasters Rapporteur: Kang Thean Shong (MMS)

Malaysian Meteorological Department Jalan Sultan 46667 PETALING JAYA Selangor, MALAYSIA

Email: kang@kjc.gov.my Fax: 603-79550964 Working Group: Peter Baddiley, Reggina Cabrera, Xu Jing Abstract: When a tropical cyclone moves over land, it generates (i) storm surges and heavy rain that result in severe flooding, landslides and debris flow; (ii) strong winds that uproot trees and damage vehicles, buildings etc and (iii) tornadoes activities that cause tremendous damages. The severity and extents of the hydrological related disasters, particularly flooding, landslides and debris flow, caused by tropical cyclone appear to increase in recent years. Thus an early warning system for hydrological related hazards is very crucial to ensure that timely and appropriate preventive and mitigation actions can be taken to reduce loss of lives and socio-economic damages. This report will summarize the operational forecasting tools and models for flooding, landslides and debris flows used in a number of countries. 0.5.1 Introduction The hydrological related disasters caused by tropical cyclone are influenced by many natural phenomena and human activities. Among them, rainfall is a major triggering factor. The main problems encountered in the early warning systems for tropical cyclone induced hydrological hazards are forecasting the pattern and rate of rainfall produced by the storm and the storm’s track. The other problem is the understanding of the interaction between the rainfall patterns with topography and environmental flow features. Practically all nations affected by hydrological related disasters, whether caused by tropical cyclone or not, have an early warning system in place. Most nations maintain at least a minimum essential observation station network, including automatic weather stations, upper air and radar stations, meteorological satellites data receiving and processing facilities to monitor the weather development and river and tidal gauges to monitor water levels and tides and wave conditions. But it is more critical and essential to enhance hydrological related disaster early warning systems by identifying effective decision-support tools and best practices. Some of the essential components of early warning systems for hydrological related disasters are flood, landslide and debris flow hazard mappings, flash flood and sediment disaster forecasting and warning, flood forecasting model evaluation and reservoir operations for flood management. Tropical cyclones take away many lives, damage properties and severely disrupt the socio-economic activities of many countries whenever they make landfall. Realizing the catastrophic impacts of tropical cyclone on socio-economic activities of a nation, the World Meteorological Organization (WMO) helped many nations to enhance their capabilities in related disaster management, including flood forecasting

57

and warning. WMO through its six Regional Specialized Meteorological Centres (RSMCs) with specialization in tropical cyclones, also issues forecasts and advisories around the clock on the occurrences of tropical cyclones. 0.5.2 Forecasting Tools and Models For Floods For operational flood forecasting, China uses various mathematical hydrological models, such as watershed hydrological models used in Xinanjing and Shaanbei (runoff yield under excess infiltration), API model, Sacramento model, Synthesized Constrained Linear System, channel routing models which used either Maskingum routing method or Lag/K routing method or Linear Diffusion Wave Routing method or Dynamic Wave routing method, Empirical methods such as P+Pa~R and relation curve of water level between upstream and downstream, and commercial model package such as MIKE 11. Some of the application of GIS and DEM in hydrological forecasting includes, delineating basin boundary and estimating basin area, generating Thiessen polygons of rain-gage network, calculating areal-mean rainfall, computing watershed parameters like land slope and river. China also planned to use the radar rainfall products in real time flood forecasting. The Bureau of Meteorology (BOM), Australia does not run hydraulic model, but the results of the hydraulic model are nevertheless used to refine the operational flood hydrological model, that is a distributed network storage routing model (named URBS). A typical configuration for the tropical cyclone-affected river systems along the Queensland coast has various models or routines that generates and simulates parameters or products such as (i) temporal and spatial variability of rainfall and runoff; rainfall-sub-catchment routing; dam storage routing and stream routing; (ii) non-linear storage-discharge relationship for sub-catchment routing with sub-areas typically resolved to about 50 to 100 km2; (iii) Muskingum routing of channel flows; (iv) dynamic calculation of each sub-area rainfall using the observed rainfall inputs from available (serviceable) stations; (v) simple rainfall loss (initial loss, continuing loss or proportional runoff, loss recovery and saturation); (vi) optional real-time forcing of observed internal hydrographs and downstream boundary conditions; (vii) multiple forecast locations within each river basin (with output of both discharge and height hydrographs); (viii) forecast rainfall with several options for spatial and temporal distribution; and (ix) observed and forecast downstream boundary condition, including astronomical tide and storm surge/tide prediction. South Korea implements the Unified Flood Forecasting System(UFFS) that integrates the existing five regional flood forecasting systems of the country. The hydrological models used in the system include the Storage-Function method, init hydrograph method, kinematic method, and a hydraulic model, DWOPER (Fread, 1987), developed by US NWS. An integrated operation system of reservoirs and rivers for flood control, including rainfall-runoff model, hydrological channel model, optimal joint operation model for reservoir system, simulation model for reservoir system and hydraulic channel routing model was also developed. • In Japan, it is compulsory for authorities responsible for areas with medium- and small-sized rivers to designate anticipated inundation area for which flood hazard maps have to be prepared by municipalities concerned. Whereas for authorities responsible for areas with large rivers to provide warnings of expected inundation areas and water levels during flooding. In the United States of America, National Weather Service River Forecast System (NWSRFS) uses either hydrologic or hydraulic models to forecast the water levels for America’s rivers and streams at about 4,000 locations. If the watershed involved in the flood forecast is located near the coast, the dynamic routing techniques and hydraulic models are used to address the effects of tides and storm surge. The use of estuarine models that include the coupling between meteorological-hydrological-hydraulic-estuarine-tide models is also becoming a part of the forecast during normal and stressed conditions. These prototypes would include the forecasting of storm surge during hurricane conditions. In USA the depiction of the inundation area, is also provided to the end

58

user and emergency personnel though a simpler technique, called Simplified Hydraulic Routing Technique (SHRT) is being proposed. A simplistic hydrological models, based either in antecedent precipitation indexes (API) or initial Soil Moisture conditions, are developed to forecast flash flood events for basins and catchments with short lead times. The U.S. Geologic Survey (USGS) has also developed a hydrological model, Precipitation-Runoff Modeling System (PRMS) that can be run under daily time steps as well as under storm events for temporal resolution of less than one hour. The model is a modular-design, deterministic, distributed-parameter modeling system developed to evaluate the impacts of various combinations of precipitation, climate, and land use on stream flow, sediment yields, and general basin hydrology. 0.5.2.1. Limitations, Challenges and Issues Forecasting of flood caused by land falling tropical cyclone still contains much uncertainty due to a number of constraints. Many times the adequacy of the model is determined but the developing hydrologic and/or hydraulic data and precipitation network are non-existent or poor to ensure good calibration of the model. The forecaster must also be experience to use the appropriate model, under the circumstances he/she is facing. The important issue is that he/she recognizes that he/she must address differently coastal, slow responding areas to flashy, steep, fast responding basin. Some of pertinent observations and issues, all associated with the limitations of quantitative precipitation forecasting, of the meteorological inputs to the hydrologic modeling process that need to be further enhanced are (i) flood-producing potential of approaching tropical cyclone; (ii) forecast or actual landfall location; (iii) quantitative precipitation forecast (QPF) which is the most important input, possibly by at least an order of magnitude; (iv) radar rainfall estimation and (v) storm tide prediction. The spatial resolution of gauges and radar, plus the incorporation of the dense networks is also critical. This will help in the calibration of radars and also in the determination of precipitation estimates based on gauges only. 0.5.2.2 Future Direction There is a need to continue in improving operational hydrologic and hydraulic flood modeling, and this includes consideration of a tighter coupling with meteorological, hydrological/hydraulic, and tidal modeling outputs where useful. We need to have a better definition of the track and intensity forecast of tropical cyclone path and quantitative precipitation forecast (QPF) associated to tropical models. The use of ensemble forecast and short term QPF should be considered. Some other works that need to be considered include;- (i) implementation of continuous soil moisture accounting models; (ii) revision of current rainfall inputs to hydrologic models to allow for gridded rainfall inputs from improved operational rainfall spatial analysis and forecast rainfall grids from NWP models; (iii) digital representation of hydrologic model sub-areas to enable improved spatial rainfall inputs (iv) use of radar-rainfall estimation, and forecasting for short lead times, for flash flood situations, (v) hydraulic flood plain modeling for continued refinement of hydrologic models, and for possible operational use in defined areas, (vi) review the methodology to generate inundation maps and the data needed for a product to accurate depict what the user needs, in terms of resolution and (vii) Increase the observation network, frequency, type, etc. and to obtain bathymetry at better spatial resolution and flows and stages at smaller time steps. Intense rainstorms brought about by tropical cyclones often caused severe landslides and debris flows, which have claimed many lives and properties. It is imperative that the flood forecasting agencies do research into and develop forecasting models for landslides and debris flows. Currently this is an area where the USGS and NWS are partners to develop such forecasting.

59

0.5.3 Forecasting Tools and Models For Geological Disasters (GDs) In Japan, one of the methods used to forecast sediment disaster is The simplified critical line setting method. Japan is revising its sediment-related Disaster Prevention Law to make it compulsory for authorities to provide sediment-related disaster hazard maps and information with risk predictions. Korea uses an algorithm called SINMAP to predict possible landslide sites. The effects and proper ranges of three calibration parameters of SINMAP, i.e. soil friction angle, cohesion and T/R could be used as an effective screening tool for landslide hazard mapping especially for mountain areas with fairly steep slopes and relatively thin soil layers. Landslides and debris flows induced by local storms or persistent rainfall are the most frequent geological disasters (GDs) all over China. In China, the basic simple approach of geological disasters (GDs), landslides and debris flows, forecast is to investigate the statistical relation between historical GD occurrences and rainfall observations so as to determine the criteria for rainfall that can trigger GDs. Five (5) rainfall-related factors (i.e. the same day rainfall, the previous day's rainfall, the accumulative previous days' rainfall, the previous wet days and the effective rainfall) are selected to build the GD occurrences model. For each factor, a set of thresholds and their corresponding probabilities of GD occurrences are determined. The overall probability of GD occurrence at a site is the weighted sum of these individual probabilities. A GD occurrence model was built for many sub-areas according to the geology and the climate of China and a national map for the spatial distribution of GD probabilities was produced. Finally, probabilities of 10%, 25%, 50%, 75% and 95% are selected to divide the results into 5 levels, from I to V with increasing susceptibility to GDs. The final forecast product is a national GD risk map distinguished by 5 levels. The GD forecast would be published through mass media if a level IV appears in an area (at least covering three weather stations), while a GD warning will be made when a level V is forecast. China is also developing an advanced GD occurrences model using Logistic regression In brief, if we denote by ps,t the probability of GD at site s and on the day t in the data set, conditional on a predictor vector xs,t, then the model is given by

ln ( ps,t / (1 – ps,t )) = X’s,tβ for some coefficient vector β. A wide range of elements has been selected as predictors. GD occurrences data and daily rain observations can be regarded as the dynamic part of the model. The relatively 'static' part, including topographical, geological and land use information was remotely sensed and constitutes the background of GD occurrences. This model has showed satisfactory performance in practice and will soon be replacing the current basic simple approach model. 0.5.3.1 Limitations, Challenges and Issues Landslides and debris flows are local disasters in mountainous areas with horizontal scale generally less than 1 km. These small-scale GD phenomena pose a very serious challenge for their prediction and prevention. Thus the difference of spatial scales between GDs and meteorological variables should be considered carefully when predicting GDs using precipitation information. In addition, accurate GD forecast requires real-time observations of extensive geological conditions and accurate predictions of rainfall. It is difficult to accurately predict the occurrences of GDs at a high spatial resolution under current circumstances. Currently available GD forecast models are only for probability predictions at low spatial-temporal resolutions.

60

0.5.3.2 Future Direction There are several issues to be considered to build an effective forecast and warning system for landslides and debris flow hazards especially for country with diverse natural conditions as big as China. There is a need to have a denser nationwide GD observational network. A closer joint researches and regularized collaborations on study of GDs, including tropical-induced GDs, by scientists from different disciplines and among different governmental agencies also should be in placed. 0.5.4 Summary Though some of the existing models in use in operational hydrological related disasters forecasting are highly simplified conceptual representation of rainfall-flood and rainfall-geological disaster response and fail to model the complexities of the land-based water cycle, it was found that the simplified models provide reasonably good results. It is very difficult to forecast the hydrological related disasters and forecaster should rely on personal experience in many parts of forecasting. There is an urgent need to improve the capacity of meteorological and hydrological services so as to jointly deliver timely and more accurate forecast and warning for disaster managers. References Dai, F. C., Lee, C. F., and Wang, S. J. (1999). Analysis of rainstorm-induced slide-debris flows on natural terrain of Lantau Island, Hong Kong. Engineering Geology, 51:279-290. Dobson, A. (2001). An Introduction to Generalized Linear Models (second edition). Chapman and Hall, London. Jiang, X. W., Wen, L. M., and Zhang, J. (2000). Experiment research of debris flow's critical rainfall based on the Langlie method. Journal of Xi'an Engineering University, 22(4):61-64. Li, Y., Meng, H., Dong, Y., and Hu, S. E. (2004). Main types and characterisitics of geo-hazard in china | based on the results of geo-hazard survey in 290 counties. The Chinese Journal of Geological Hazard and Control, 15(2):29-34. Lin, M. L. and Jeng, F. S. (2000). Characteristics of hazards induced by extremely heavy rainfall in central taiwan | Typhoon Herb. Engineering Geology, 58:191-207. Liu, Y. (1998). The criterion of rainstorm intensity associated with landslides. Hydrogeology and Engineering Geology, 3:43-45. Tan, W. P., Luo, X. M., and Wang, C. H. (2000). Forecast models of rainstorm debris flows.Journal of Natural Disasters, 9(3):106-111. Wei, Y. M. and Yu, Y. Y. (1997). Study on prediction models of precipitation-typed debris flow. Journal of Natural Disasters, 6(4):48-54. Wen, K. J., Wang, L. X., Xie, B. Y., Yu, Z. M., Lin, D. Z., and Du, P. Z. (1998). Real time forecast of storm caused debris flow. Journal of Beijing Forestry University, 20(6):59-64. Wu, J. S. (1990). Observational study of debris flows in Jianjiagou, Yunnan Province. Science Press, Beijing.

61

Zhu, L. F., Wu, X. C., Yin, K. L., and Liu, X. G. (2004). Risk zonation of landslide in China based on information content model. Journal of Earth Sciences and Environment, 26(3):52-56. Country Reports by The People’s Republic of China, Japan and Republic of Korea during the 38th Session of ESCAP/WMO Typhoon Committee, 14-19 November 2005, Hanoi, Vietnam

62

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES

Topic 1 : Tropical Cyclone Structure and Structure Change Topic Chair: Hugh E. Willoughby

Florida International University Department of Earth Sciences, PC 344 University Park Campus Miami, FL 33199

E-mail: hugh.willoughby@fiu.edu Fax: 305.348.3877 Rapporteurs: E. A. Ritchie, J. Kepert, L. K. (Nick) Shay, M. A. Lander, and J. Knaff Abstract: Tropical-cyclone (TC) structure and structure change, including intensification and weakening, is at the frontier of forecasting science. Knowledge and techniques within each sub-topic here have seen significant advancement during the last decade, much of it during the four years since IWTC-5. A more detailed understanding of the inhibitory role of shear of the prevailing wind around TC has emerged, and it has become evident that not shear alone, but shear in combination with environmental middle-level humidity and the strength of the oceanic enthalpy source dominates the process. Significant advances have occurred in understanding aspects of TC internal dynamics, including classical (and not-so-classical) “hot towers,” spiral bands, concentric eyewalls and eyewall replacements, vortex Rossby waves, boundary-layer dynamics, and the eye as a source of high Eθ air. Increasingly powerful computers have opened the possibility of numerical simulation with non-hydrostatic numerical models at resolutions of 1-2 km to represent these features realistically for individual TCs. Observations from spaceborne radar altimeters combined with those from air deployable probes clearly define strength of the oceanic moist enthalpy source in relation to deep oceanic mixed layers and strong local currents. Representation of these effects using a new generation of high-resolution, coupled models is beginning to improve intensity forecasts. In the Atlantic Basin, piloted heavy aircraft provide incomparable data for model initialization and ground truth for remote sensor development. Autonomous aircraft promise a similar, though less elaborate, capability in other basins. Nonetheless most forecasters, most of the time, will continue to rely upon spaceborne remote sensing. Although the Dvorak technique has matured and is now universally applied, recent advances in TC structure and intensity observation focus on spaceborne active and passive microwave sensors, generally deployed on satellites in steeply inclined orbits. Sensitivity of meteorological motions to initial conditions means that predictions from future sophisticated numerical models will be communicated in probabilistic terms. Portraying uncertainty to everyone from sophisticated decision makers to the general public will be a continuing challenge, as will validation of probabilistic forecasts. 1.0.0 Introduction TCs are relatively small (horizontal scale ~1000 km), discrete vortices that require a few days to a month to complete their life cycles. Historically, their meso spatial scales and synoptic time scales have meant that forecasts were framed in terms of TC parameters. By “parameters” one means quantities like latitude and longitude of the center, minimum pressure, maximum wind, radius of maximum wind,

63

and so forth. Currently, advisories are cast in terms of TC parameters to the extent that warning centers have access to observations or models that can evaluate them. Additionally, in applications that assess underwriting risk or extent of storm-surge flooding, impacts are generally evaluated using “parametric” models that convert climatological parameters into geographic distributions of weather elements. Before the late 1990s, a key obstacle to forecasting detailed impacts in real time was inability to simulate realistic structure and intensity--even if the track were correct. For example, size of simulated vortices was controlled by model resolution rather than by physical processes. Although the minimum pressure might be representative, coarse resolution caused too-weak maximum winds. Lack of coupling between the TC winds and ocean response exaggerated the oceanic heat source. Since the introduction of the GFDL model in the mid-1990s, operational models have become increasingly realistic. Demonstration experiments with high-resolution research models are extremely promising. The reality, however, is that while operational intensity forecast produced by some of these models are now skillful, the forecast of vortex wind structure is not; suggesting further improvement in vortex initialization, model resolution, and physical parameterizations are needed. Fortunately, both the well-established MM5 and new WRF models can potentially simulate hurricane structure and intensity in great detail. The forecasting community seems to be approaching a regime where uncertainty about the initial condition will predominate over model shortcomings as a limitation on forecast accuracy. Even as computational power and model sophistication have advanced, so too has understanding of TCs’ essential physics. 1.0.1 External Atmospheric Forcing Shear of the surrounding, synoptic-scale wind is the dominant atmospheric forcing agent for TC structure. If it is imposed suddenly in a model, it causes the vortex to precess cylonically and then assume a down-shear tilt. In both models and observations, convective cells trigger on the downshear side of the vortex, advect cyclonically around the left-of-shear side (northern hemisphere), and appear to dissipate on the upshear side. In actuality, buoyant bubbles generally rise above the 0oC isotherm by the time they reach the upshear side. When the suspended hydrometeors freeze or fall out, the reflectivity decreases, but the updrafts continue to rise toward the tropopause as they are carried the rest of the way around the center to the downshear side. Forecasters recognize shear-induced asymmetry easily in satellite and radar images. Since shear’s negative influence in intensity is well known, diagnosis of shear patterns is a key element of satellite intensity estimates. Shear does not act independently of other environmental influences. The earliest hypothetical mechanism for shear interaction was “ventilation,” advection of cooler environmental air into the warm core of the TC vortex. Although it is possible to envision interactions involving eddy exports of angular momentum, for example, something very much like ventilation actually seems to occur, but it is in the lower troposphere, where the diabatic inflow takes place rather than in the upper troposphere where the temperature anomaly responsible for the low hydrostatic pressures lies. Entry of environmental air at 700 hPa, near the θE minimum, supplies low-moist-enthalpy air that promotes cold downdrafts leading to lower buoyancy at the base of convection. Thus, shear combines with a dry environment to inhibit intensification. Observations of Atlantic TCs interacting with the Saharan Air Layer (SAL) confirm this connection. Other differential properties of the flow around TCs can also force interactions. Environmental deformation excites wavenumber-two asymmetries, and environmental vorticity gradients can interact to force wavenumber-one gyres similar to those forced by the poleward increase of planetary vorticity.

64

1.0.2 Inner-Core Dynamics The thermodynamics of TC intensification resemble a classical Carnot heat engine in which the Maximum Potential Intensity (MPI) is determined by the temperatures of a warm heat reservoir at the Sea-Surface Temperature (SST, Ts, ~300K) and a cold reservoir at the tropopause temperature (Tc,~200K). MPI is proportional to the thermodynamic efficiency (Ts – Tc)/Ts. Inclusion of the heat generated by frictional heating at the ocean surface leads to enhanced efficiency in which Tc replaces Ts in the denominator. Most TCs fail to attain MPI. In somewhat more detail, the Sawyer-Eliassen process (SEP) describes the kinematics of TC intensi-fication. Latent heat release in the convective updrafts around the eye lofts air from the top of the frictional boundary layer to the tropopause. The updrafts entrain middle-level air as they rise. Thus, frictionally controlled inflow supplies moist enthalpy to sustain the updrafts and the diabatically induced middle level inflow supplies angular momentum needed to spin up the vortex. Intensification through this process is manifested primarily as an increase of the maximum wind and contraction of the radius of maximum wind. A reasonable, but as yet unproven, conjecture is that rapid intensification occurs when the SEP is unfettered by shear, dry-air ventilation, or storm-induced oceanic cooling. Nearly all major (vmax > 50 m s-1) TCs form through rapid intensification. The SEP is quasi-steady in the sense that characteristic times (i.e. period of sinusoidal forcing) are longer than the local inertia period, − −π ∂ ∂ + +1 12 [( ( )/ )(2 / )]r rv r f v r f . The primary effect of time varying or asymmetric heating is to induce the same change as a quasi-steady symmetric injection of the same amount of net heat. Since the early 1980s, it has been recognized that TCs, particularly intense ones, may exhibit multiple concentric eyewalls. Often the outer most eyewall, which intercepts the high θE inflow, will contract and supplant the inner eyewall, leading to a temporary weakening of the TC as a whole. Experience indicates that if conditions (low-shear, warm SST) remain favorable, the TC will generally re-intensify. If not, the eyewall replacement often marks the beginning of TCs’ final weakening. Sometimes the eyewall replacement results in formation of an “annular hurricane” composed of an isolated eyewall without spiral band or outer rings of convection. These cyclones do not undergo further eyewall replacements. They generally have intensities about 85% of MPI and seem to be more resistant to shear. Prediction of the timing and intensity modulation due to eyewall replacements remains problematic, although recent results from high-resolution numerical simulations are encouraging. Although the foregoing description seems to deprecate the roles of non-symmetric, non-steady processes, a well formulated theory of vortex asymmetries has emerged. It encompasses both inertia-buoyancy and vortex Rossby waves (VRWs). In common with their synoptic-scale analogs, VRWs transport wave energy, momentum, and enthalpy. For example, observations show that the air in contact with the sea in the lowest ~kilometer of the eye can attain the highest values of θE anywhere in the TC. In rapidly intensifying TCs, the wind profile inside the eye becomes U-shaped in response to the SEP. Thus, the axially symmetric relative vorticity has a maximum between the eyewall and the center of rotation. It then meets the necessary criterion for barotropic instability. Consequently, counter-rotating trains of growing VRWs form, straddling the vorticity maximum. These waves also mix high θE air into the eyewall updrafts to “supercharge” the SEP. VRWs may play other significant roles in TC intensification and motion. Dropsonde observations of the TC boundary layer show a wind maximum at ~500 m altitude in the vortex core and somewhat higher, ~ 1 km, farther from the center. The profile-maximum wind in this jet tends to be 10-30% stronger than the wind at 2-3 km altitude where reconnaissance airplanes fly. The surface wind is somewhat weaker, 80-95% of flight-level wind. Beneath the eyewalls of intense TCs, or on the left of the motion (right in the Southern Hemisphere) it may be 100% of flight-level winds. The strength of the jet at 300-800 m increases toward the TC center. Deceleration of the inflow followed by outward acceleration as air rises above the frictional boundary layer is the primary cause of the jet.

65

Models and observations show that the greatest inflow occurs near the front of the storm, with the strongest jet about 90 degrees of azimuth downstream on the left side of the motion vector(right in the Southern Hemisphere). 1.0.3 Air-Sea Interface and Oceanic Influences The sea surface provides the warm reservoir for MPI and other thermodynamic calculations, but pre-existent SST is not a good measure of the actual thermodynamic boundary condition at z = 0. As they pass, TCs induce mixing at 1 to 3 times the radius of maximum winds and upwelling in the ocean along the track. Shallow mixed layers and slow TC motion increase this effect; deep mixed layers and moderate to fast moving TCs decrease it. Because water has a specific heat twice a great as dry air and is 800 times denser, a few meters of water have the same heat capacity as the entire tropospheric column. Thus, most of the cooling due to TC passage results from entrainment of cold water across the thermocline with only a small contribution from the energy extracted by the storm. As shown above, SST that remains in the range 27-28oC is essential to maintaining the most intense TCs.

During the 2005 Atlantic Hurricane Season, four devastating hurricanes, Dennis, Katrina, Rita, and Wilma intensified over the Gulf of Mexico. The first three reached greatest intensity as they passed over the Gulf of Mexico Loop Current (Fig. 1.0.1). They weakened subsequently over the shallower mixed layers in the Gulf Common Water between the Loop Current and landfall. Wilma was a late-season hurricane that reached greatest intensity over the very high ocean heat content water of the western Caribbean, weakened over the Yucatan, and then reintensified as it moved northeastward

Fig. 1.0.1 Maximum wind (open circles)and minimum pressure (filled squares) histories of Hurricane Rita as it traversed the Gulf of Mexico, illustrating the role of the Loop Current in Rita’s rapid intensification.

66

while in substantial environmental wind shear across the Loop Current to landfall in Florida as a Category 3 hurricane. These events, and the increasing ability of fully coupled models to simulate them, emphasize the need for models that can be initialized to represent both warm and cold mesoscale oceanic features usually associated with strong currents. Another aspect of the modeling is to capture the extent of the upper ocean cooling induced by shear-induced mixing at the base of the ocean mixed layer. Internal-wave driven shear (primarily at the inertial frequency) lowers the Richardson number below criticality and forces the ocean mixed layer to deepen and cool. However, the extent of the cooling is a function of the strength of stratification, initial layer depth and background currents. Thus, in a broader context, the levels of low-frequency internal wave shear in the upper ocean are analogous to the levels of shear in the atmosphere in affecting TC intensity. A second frontier of investigation is turbulent transport across the ocean surface. Dropsonde wind profile measurements near the surface show that the wind profile is logarithmic up to 200 m altitude. Surface roughness and drag coefficients deduced from these profile decrease for winds > 33 m s-1 so that the drag itself increases very slowly with increasing wind. Laboratory experimentation as well as oceanic response studies find a similar leveling off the surface drag coefficient between 28 to 33 m s-1. It remains unclear whether the surface drag coefficient decreases after this threshold value. Similarly, the enthalpy coefficient, which combines sensible and latent heat exchange effects, appears to level off beyond this wind speed threshold. Observations also support sea-state dependence of these surface exchange coefficients, however wind-wave effects only modulate intensity fluctuations rather than being a direct cause of large intensity and structure changes, as suggested by recent coupled modeling experiments. Nonetheless, these relationships point to a need for three-way coupling between atmospheric, ocean thermal structure and currents, and sea state in the next generation of TC models. 1.0.4 Operational Techniques for Defining Structure

Characterization of TC structure and intensity has always been challenging because observations over the sea are so sparse. Starting in the middle 1940s, aircraft observations began to address this problem in the Atlantic and northwestern Pacific. They provided position, maximum estimated surface wind, and extrapolated central pressure. Initially great faith was placed in pressure-wind relations. Airborne and land-based radar provided fairly reliable estimates of radius of maximum wind. The only other measures of structure were manually recorded estimates of outer winds from reconnaissance aircraft and sparse ship, coastal, or island reports.

The advent of polar orbiting provided more position fixes. In the mid-1970s, geostationary satellites supported intensity estimates based upon the Dvorak technique. At approximately the same time, air-craft Inertia Navigation Equipment (INE) allowed accurate measurement of flight-level winds. Aircraft observations could be transmitted to warning centers on low-bandwidth Aircraft-Satellite Data Link (ASDL). In the early 1980s, airborne active C-Band Scatterometers and passive Stepped-Frequency Radiometers promised remotely sensed surface winds, although calibration issues were problematic at first. By the middle 1980s, routine aircraft reconnaissance was confined to the Atlantic with occasional deployments to the northeastern and central Pacific. Thus, the modern climatology of detailed TC structure and intensity is derived from those basins.

Starting in the mid-1950s analytical approximations to the wind as a function of radius began to appear. The earliest version had a single spatial scale, defined by the radius of maximum wind, and a single velocity scale defined by the radius of maximum wind. If the structure of real TCs were actually defined so simply, a one-to-one relationship between minimum pressure and maximum wind would exist, apart from a small latitude-dependant correction proportional to the inverse Rossby-number at the radius of maximum wind.

67

At the turn of the 21st century, spaceborne sensors--principally scatterometers, passive microwave radiometers, and active radars--seem to offer means to determine TC structure globally by remote sensing. Scattermometers (i.e. QuikSCAT) can sense outer wind structure, although rain contamination and large footprint problems preclude observation of the vortex core. Passive radiometers (i.e. AMSU) can retrieve soundings in the TC environment and estimate surface pressures, at least in geometrically large TCs. Radiometers operating in the 85 GHz band produce images that mimic surface PPI radars. They, along with the TRMM spaceborne radar, can define TC convective structure and rain rate. Although development of algorithms to retrieve intensity by deconvolution of coarse-resolution sounding data given accurate estimates of TC structure have not been forthcoming, this approach seems feasible.

1.0.5 Operational Guidance and Skill

As track forecasts have improved, users have come to expect more information on structure and intensity. The concept of “intensity,” as measured by minimum sea-level pressure or maximum wind, is to a great extent artificial. It presents a decision maker with a worst-case estimate of the strongest wind if the storm were to pass directly over a given station. Despite significant advances in numerical weather prediction, the best intensity forecasting guidance comes from statistical models, such as the extrapolation built into the Dvorak technique or the SHIPS statistical technique, and specialized hurricane models, such as GFDL and GFDN. In recent years however, these guidance methods are driving very slow operational intensity forecasting improvements.

Short-term improvements of operational intensity forecasts will likely be realized by using consensus and ensemble methods formed from skillful and independent intensity guidance methods. In the long-term however, intensity forecast improvements will likely come from advanced operational numerical forecast systems. These models will be coupled with the ocean, including wave influences, will incorporate advanced data assimilation techniques which make use of all remotely sensed data, and explicitly resolve convection.

Decision makers require information on the onset and extent of gale-force, 50-kt, or hurricane-force winds. Current guidance based upon imperfect observations and statistical models is, at best, marginally skillful and forecasts provided by current operational numerical prediction models are not skillful. In many circumstances, the geographical extent of overcast or of radar reflectivity serves as a proxy for quantitative forecasts of weather elements. If ensemble forecasting with realistically intitialized high resolution numerical models proves to be feasible, it may be possible to forecast geographical distributions exceedance probabilities for crucial TC impacts. Although these forecasts will prove difficult to verify, they will also support rational decision making with well-defined cost-benefit ratios for preparations. 1.0.6 Summary

Accurate forecasts of TC structure and intensity lie on the scientific frontier. Essential prerequisites include: high-resolution (< 2 km) numerical models, ability to initialize them (for most of the world using remotely-sensed data); accurate coupling with ocean thermal structure, current and sea state; and improved representation of physical processes, such as microphyics and turbulent mixing. Because mesoscale motions are inherently sensitive to initial conditions, usable forecasts will probably require ensembles of these models. The benefit of this capability will be accurate and precise forecasts of geographically distributed weather elements, albeit cast in stochastic terms.

68

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP ON TROPICAL CYCLONES

Topic 1.1 : Environmental Effects on Tropical Cyclone Structure and Structure Change Rapporteur: E. A. Ritchie Department of Atmospheric Sciences PO Box 85721-0081 University of Arizona Tucson AZ 85721-0081 Email: ritchie@atmo.arizona.edu Fax: 520-621-6833 Working Group: J. Dunion, M. Guishard, S. Jones, S. Kimball, C-S Liou, Y. Wang Abstract: Recent research to increase understanding, and techniques to improve forecasts, of the structure and structural changes of a tropical cyclone due to interaction with the environment are summarized. The atmospheric environment is considered here, and the oceanic, and air-sea interface environments are summarized in Topic 1.3. Progress in understanding how a tropical cyclone interacts with its environment, and in developing techniques to forecast tropical cyclone intensity- and structure-change events, has been made over the past few years. Whereas any increase in skillful forecasts due to these research and techniques may take some years to develop, it is felt that improvement in tropical cyclone intensity forecasts is likely, due in part to the work described here. 1.1.1 Introduction The impacts of the environment on tropical cyclone structure and structure change have been studied for many years. Here, an update on progress in research and forecast techniques since the fifth IWTC is provided. It is well known that favorable environmental conditions (including minimum vertical wind shear) are required for tropical cyclone formation. Emanuel (1988) and Holland (1997) have developed separate relationships between the maximum potential intensity (MPI) and the sea-surface temperature (SST) and the environmental conditions, which include the static stability, upper-tropospheric conditions, and relative humidity. The wind structure (intensity) changes of a tropical cyclone from formation to maximum intensity to decay depend on a balance between favorable and inhibiting environmental conditions. Whereas, mostly atmospheric factors will be considered here, in a later topic (1.3) the sea-surface and oceanic forcing will be summarized. Environmental conditions summarized in the following sections include: 1) Low- or No-flow environments – the wind field is near zero throughout the troposphere; 2) Uniform flow environments – the wind field is near constant throughout the troposphere; 3) Vertical wind shear environments –the mean wind changes with height. The most common measure of vertical wind shear is the mean wind at 850 hPa subtracted from the mean wind at 200 hPa although different definitions do exist. The resulting value has both a magnitude and a direction. Furthermore, a shallow shear is sometimes defined as the mean wind at 500 hPa minus the mean wind at 850 or 925 hPa. The mean wind is usually calculated as an average over an area extending

69

radially from the center of the tropical cyclone. Mean winds have been calculated over a circular area extending up to 600 km from the tropical cyclone center, or in annular areas extending, for example, from 200 km radius to 600 km radius, or 200 km to 800 km radius from the center of the tropical cyclone. Whereas vertical wind shear is strictly a measure of velocity gradient with units of either s-1 or m s-1 (hPa)-1, it is more commonly reported as m s-1 with the depth over which the shear is calculated implied as 200 – 850 hPa or stated. There is no standardized measure of shear as a function height. i.e., a simple subtraction of the winds at two levels does not represent the vertical structure of the shear. It is possible to have the same shear calculated between 200 and 850 hPa when in one case, the shear is all located in the upper atmosphere, and in another, in the lower atmosphere. It is expected that the tropical cyclone would respond differently to these two types of mean vertical wind structure; 4) Upper-level trough interactions; 5) Environmental moisture and the Saharan air layer. 1.1.2 Low- or no-flow environments

A study by Knaff et al. (2003a) highlights a small subset of tropical cyclones in the North Atlantic and eastern North Pacific basin that briefly developed unusual structural and intensity characteristics in low easterly vertical wind shear environments over constant or decreasing SSTs. As observed in infrared imagery, these tropical cyclones tended to have larger than average eye sizes, symmetrically distributed cold brightness temperatures in the eyewall, and little or no rainband features. In addition, these “annular” tropical cyclones were significantly stronger, maintained their peak intensities longer, and filled more slowly, than the average tropical cyclone in these basins. Furthermore, average official forecast intensity errors for these types of tropical cyclones were 10 – 30 % larger than the 5-y mean official errors during the same period. A simulation of a tropical cyclone in environmental conditions similar to those found by Knaff et al. (2003a) to be conducive to annular-hurricane formation was more intense than its beta-plane counterpart, and had a more axisymmetrically distributed inner-core precipitation pattern similar to that inferred from infrared imagery for the annular hurricanes, and more power in the wave-number-0 (symmetric) component of the potential vorticity field (Ritchie 2004). As a personal observation, it has been interesting to note the use of the term “annular” in TC-community discussions of existing tropical cyclones. It seems that this research into one class of “outliers” has begun to pay off.

An idealized modeling study by Ritchie and Frank (2006a) found that a tropical cyclone on a f-plane (i.e., no large-scale asymmetries) could develop vertical wind shear within at least 300 km of its center due to its own unstable outflow (Wong and Chan 2004). This weak vertical wind shear (maximum of 4.5 m s-1 averaged over the inner 300-km radius) was strong enough to generate small-scale asymmetries in convective and precipitation structure. It was found that if shear of magnitude greater than 2.3 m s-1 persisted for more than about 3 hours, then persistent, shear-forced asymmetries in the convection and precipitation fields would develop. Note that these asymmetries in the structure of the TC did not appear to impact the rate of intensification of the tropical cyclone to any degree. 1.1.3 Uniform (non-zero) Flow It would be unusual for the near-tropical cyclone environment (within 1000 km radius) to consist of a uniform flow because of the vertical wind shear associated with the beta gyres. However, it is educational to consider how a mean flow on a f plane and a beta plane affect tropical cyclone structure and intensity. A previous study (Frank and Ritchie 2001) found that a weak uniform (3.5 m s-1) background flow that intensified slightly more rapidly, and reached a slightly higher intensity, than the comparable control no-flow case. In this case, the asymmetry in convection produced by frictional convergence in the front quadrant of the tropical cyclone (Shapiro 1983) produced an enhancement in

70

the average precipitation of 2-5 cm (3 h)-1 in the inner 50 km of the tropical cyclone when compared to the no-flow case (Frank and Ritchie 2001).

A more recent study by Kwok and Chan (2005) found that a strong uniform flow (6-8 m s-1on a f plane) resulted in a weakening of the tropical cyclone. Their principle finding was that the interaction between the uniform flow and the tropical cyclone circulation resulted in a wave number one asymmetry in vertical motion, which, along with a rotation of the upper-level anticyclone, produced an effective vertical wind shear over the tropical cyclone. In weak flow cases (0-4 m s-1), the vertical motion asymmetry was not induced, and no reduction in intensity was observed, a result consistent with the earlier finding of Frank and Ritchie (2001).

The direction of environmental uniform flow has also been found to be a factor in modeling studies when the beta effect is included. Peng et al. (1999) found that uniform westerly flow was more favorable for tropical cyclone intensification than uniform easterly flow. They concluded that westerly (easterly) uniform flow partially cancelled (enhanced) the northwesterly tropical cyclone motion induced by the beta gyres and thus reduced (increased) any motion asymmetry resulting in more symmetrically (asymmetrically) organized convection. Kwok and Chan (2005) support this finding. They find that on a variable-f geometry, westerly uniform flow partially cancels the beta-induced vertical wind shear, while easterly uniform flow enhances it. This result is also consistent with Ritchie (2004) and Ritchie and Frank (2006b). 1.1.4 Environmental vertical wind shear The effects of vertical wind shear on the intensity and to a lesser extent the structure, of a tropical cyclone is qualitatively well known. Strong vertical wind shear has an inhibiting, and even weakening, effect on tropical cyclone intensification. In strong shear, the low-level center of the tropical cyclone will often become exposed, with the convection and cloud shield shifted downshear of the exposed center and the tropical cyclone will typically weaken. This result was incorporated into the original Dvorak technique to infer intensity trends from infrared and visible satellite imagery. However, the relationship between weak to moderate shear and tropical cyclone structure and intensity change is less clear. More recent studies have begun to elucidate more details regarding the effects of different strength, and structure, of vertical wind shear and how this affects the intensity and structure of tropical cyclones of different intensities.

a) Theoretical Studies – dry vortices

Recent theoretical calculations show that dry vortices exhibit resiliency to the presence of vertical wind shear (Reasor et al. 2004; Jones 2004). Mechanisms discussed include rotation of a tilted vortex such that when the vortex is tilted “upshear” the vertical shear effectively constrains and reduces the tilt, and when it is “downshear” the environment will enhance the tilt (Jones 2004). The resiliency of the vortex to the environmental shear is dependent on the value of the Rossby penetration depth, i.e., the larger the depth, the greater the resiliency of the vortex to the shear. Reasor et al. (2004) present a different, but complimentary view on the resiliency of dry vortices to vertical shear forcing. They argue that when the vortex tilts, the asymmetries are projected onto vortex Rossby waves and damped out thus reducing the tilt of the vortex. One major difference between the two papers still seems to be the location of the forced vertical motion. Jones (2004) finds it to be left of the vortex tilt (consistent with previous work) and finds that the tilt rotates with time whereas Reasor et al. (2004) find that the tilt is steady at downshear-left (contrary to other authors) where presumably the forced positive vertical motion is located. As Jones (2004) points, perhaps the definition of vortex center used to define “tilt” is critical in determining which mechanism might be working. Another recent paper by Patra (2004) supports the earlier finding by Frank and Ritchie (1999) that latent heat release in neutral ascent in the inner-core of a tropical cyclone results in a shift of the ascent

71

pattern to downshear left. One particularly interesting result is that if the region of saturated ascent is confined to one side of the vortex, the tropical cyclone can still intensify suggesting that tropical depressions and sheared tropical storms may still be able to intensify under vertical wind shear.

b) Intensification and weakening trends in tropical cyclones

Previous observational and modeling studies have shown a relationship between the strength of the environmental vertical wind shear and the amount of weakening or intensification that occurs in a tropical cyclone (Gallina and Velden 2002; Frank and Ritchie 2001). In particular, Gallina and Velden (2002) found that for Atlantic tropical cyclones, the critical shear value where the tendency changes from intensifying to weakening tropical cyclones occurs at about 7-8 m s-1 (200 – 850 hPa) of vertical wind shear. For the western North Pacific basin this critical shear value is 9-10 m s-1 (200 – 850 hPa). More recently, Wong and Chan (2004) support these findings with a series of numerical simulations. They found that whereas tropical cyclones in weak (0-4 m s-1) shear intensified during the simulation, increasing the shear to 6-8 m s-1 resulted in a relatively weaker tropical cyclone that maintained its intensity (but did not strengthen). Furthermore, at shear values of 10 m s-1, the tropical cyclone weakened significantly. These values are very similar to those empirically deduced by Gallina and Velden (2002). In support of this, Zeng et al. (2006) find that very few tropical cyclones in the western North Pacific intensify when the environmental vertical wind shear is greater than 20 m s-1, a result that is certainly not contradictory to anything previously found.

The impact of the vertical wind shear is also sensitive to the size of the tropical cyclone (Wong and Chan 2004). A smaller tropical cyclone weakened in only 4 m s-1 of vertical wind shear, and in 6 m s-1 of vertical wind shear, dissipated by 48-h of simulation. This is also similar to the observational result of Gallina and Velden that under the same vertical shear environment, more intense tropical cyclones were impacted more slowly than weaker tropical cyclones.

Zehr (2003) used a case study of Hurricane Bertha to demonstrate that the asymmetry in infrared cloud structures can be related to the model-derived vertical wind shear. This process may prove to have utility in assessing actual environmental vertical wind shear from the infrared cloud asymmetries.

An extremely interesting result is from Emanuel et al. (2004) who report adding a crude parameterization of vertical wind shear into their “CHIPS” model, which is an axisymmetric atmospheric model coupled to an ocean model for intensity forecasting. Whereas the original model did not perform well when the tropical cyclone was embedded in relatively high shear, the addition of a parameterization for vertical wind shear dramatically improved their results in these situations. The new configuration outperformed GFDL for the 2002 Atlantic hurricane season, and performed almost as well as SHIPS out to 48 h. c) Secondary circulation, convective asymmetries, and precipitation patterns

A study by Zhang and Kieu (2005) isolates the forced secondary circulation by the vertical shear of horizontal winds from the latent heating and friction circulations associated with a simulated hurricane vortex. They find that when an environmental westerly shear is superposed with an axisymmetric balanced vortex, an anticlockwise forced secondary circulation appears across the inner-core region with the rising motion downshear and easterly sheared horizontal flows in the vertical. The resulting horizontal flows act to reduce the influence of the vertical shear inside the storm by as much as 30–40%, opposing the destructive role of the vertical shear.

Recently, model simulations that include diabatic effects (Frank and Ritchie 1999, 2001; Kimball and Evans 2002) and several observational studies (Reasor et al. 2000; Corbosiero and Molinari 2002a, 2002b; Black et al. 2002; Zehr 2003) have established and verified the existence of persistent patterns of asymmetric convection and rainfall that develop in the downshear-left quadrant of the storm. The model study of Frank and Ritchie (2001) found that the asymmetries developed due to the storm’s

72

response to imbalances caused by the shear, but differ from the prior adiabatic simulations because saturation in the eyewall leads to a different lifting mechanism. Interestingly, an observational study (Corbosiero and Molinari 2002a) that used lightning flash density in 35 tropical cyclones found that in the inner core region (< 100 km radius) the flashes occurred preferentially in the downshear left quadrant, which is consistent with the predictions of Frank and Ritchie (2001). In the outer rainbands (100–300 km of the center) the preference for the lightning was for downshear right, similar to the adiabatic model studies of Jones (2000) and Frank and Ritchie (1999). These lightning distributions were valid both over land and water, and for depression, storm and hurricane stages.

A more recent modelling study by Rogers et al. (2003) has shown that while the magnitude of the convective asymmetries and instantaneous rainrate are directly related to the magnitude of the environmental vertical wind shear, the distribution of accumulated rainfall was related also to the direction of the vertical wind shear and the storm motion. The accumulated rainfall had a more symmetric distribution across the track of the storm if the shear vector was strong and across track, but showed a distinct maximum on the left side of the storm track when the shear was weak and along track.

d) Tropical cyclone thermal structure

Probably the principle reason why a tropical cyclone eventually will weaken under the influence of strong environmental vertical wind shear is because the tropical cyclone upper-level warm core cannot be maintained at a level that will continue to support the surface low pressure. Model studies (e.g., Frank and Ritchie 2001; Ritchie and Elsberry 2001) indicate that environmental vertical wind shear would impact the tropical cyclone at the upper-levels initially, which is where the inertial stability associated with the tropical cyclone primary circulation would be a minimum. Ritchie and Elsberry (2001) simulated an initial advection downstream of the upper-level warm core of the tropical cyclone, and thus a reduction in the magnitude of the warm core aloft. This resulted in a reduction in the height of the maximum warm core, an enhancement of the warm core at lower levels (due to subsidence into the core forced by convergence between the environmental winds and the cyclonic flow of the tropical cyclone), an associated rise in the sea-level pressure, and a reduction in the cyclonic flow aloft, which further reduced the inertial stability aloft. Consequently the vortex became more susceptible to the vertical wind shear and thus more of the warm core was advected downstream. Although this negative feedback could lead to continued erosion of the deep convection and upper-tropospheric warm core, and thus finally a dissipation of the tropical cyclone, Ritchie and Elsberry (2001) found that an eventual balance between the environment and (weaker, shallower) tropical cyclone was established.

The introduction of the Advanced Microwave Sounding Unit (AMSU) has allowed the routine examination of tropical cyclone thermal structure. While the AMSU soundings lack the horizontal resolution to resolve the warm core of the tropical cyclone eye, the broader-scale warm core envelope can be measured. The strength of this broad-scale warm signature has been related to intensity (Brueske and Velden 2003) and the horizontal extent of the warm core along with an estimate of maximum intensity can be related to surface wind structure (DeMuth et al. 2003). A study using the advanced microwave sounding unit (AMSU) (Knaff et al. 2004) that analyzed the temperature anomaly in tropical cyclones in vertical wind shear found that typically as vertical wind shear increased, the warm-core vortex became shallower. These observations are consistent with the modeling results.

1.1.5. Upper-level trough interactions

It is difficult to find any reported research in the past four years since IWTC-V that investigates the interactions between upper-level troughs and tropical cyclones except as part of extratropical transition, which is not covered here. The purview of this section is to report how upper-level troughs impact tropical cyclone structure and intensity while they remain tropical in structure. Previous

73

research, which is reported in existing IWTC reports, will support the thesis that the precise manner and degree to which upper-level troughs weaken or intensify a tropical cyclone's circulation is not yet well understood. Although an upper-level trough in close proximity increases the vertical wind shear over the tropical cyclone, studies have demonstrated that complicated dynamic processes occur during the interaction between an upper-level trough and a tropical cyclone that then affect the core dynamics of the tropical cyclone in ways that are only just beginning to be investigated. A trough interaction has been defined by Hanley et al. (2001) to occur when the eddy momentum flux convergence calculated over a 300-600 km radial range is greater than 10 m s-1 d-1.

A favorable factor for intensification of a tropical cyclone has been characterized as a “good trough” interaction and two types of “good troughs” have been identified and described using composite analysis (Hanley 1999; Hanley et al. 2001). In this scenario, an upper troposphere trough becomes juxtaposed with the warm outflow from the tropical cyclone to cause: (i) a positive eddy momentum flux convergence that contributes to a cyclonic spinup of the inner vortex; and/or (ii) an enhancement of the jet streak that contributes to a larger outflow from the tropical cyclone, and consequently a spinup of the vortex (Hanley et al. 2001; Hanley 1999). Kimball and Evans (2002) note that a merger between a simulated shallow upper-level trough and tropical cyclone leads to reduced vertical wind shear from the trough over the tropical cyclone. Rapid intensification of the tropical cyclone followed in conjunction with contraction of the radius of maximum winds.

In the contrasting “bad trough” scenario, the strong winds on the leading side of an approaching upper-level trough produce a strong vertical wind shear that is concentrated in the upper portions of the troposphere over the tropical cyclone. This scenario has been observed to occur when the upper-level trough either remained relatively far from the tropical cyclone so that the effective impact on the tropical cyclone was a debilitating one associated with the vertical wind shear Hanley et al. 2001). In addition, Kimball and Evans (2002) note that in their model simulations, the deformed trough inhibited outflow on the east side of the tropical cyclone, which hampered future intensification.

Although composite and model studies of an upper-level trough and its associated vertical shear interaction have provided insight into the mechanisms of trough interaction, it has proved particularly difficult to apply these insights to individual cases. In recent years, Hurricane Opal of 1995 has become one of the most intensely studied hurricanes ever. However, the cause of the hurricane's rapid intensification over the Gulf of Mexico is still a matter of controversy. Several insightful studies (e.g., Bosart et al. 2000; Persing et al. 2002; Möller and Shapiro 2002; Shapiro and Möller 2002) used a range of techniques to elucidate the role of the upper-level trough in the intensification of Hurricane Opal. However, there was no general consensus out of these studies. Furthermore, additional studies (e.g., Hong et al. 2000 and Shay et al. 2000) suggest that the oceanic warm core ring that Opal passed over during the period of rapid intensification had significant impact on the hurricane’s heat budget and thus also impacted its intensification. Hanley (2002) has used water vapor imagery with some success to identify a tropical cyclone-trough interaction, which gives some hope that continued investment in remote sensing technology may help with this forecast challenge.

1.1.6 Effects of environmental moisture (or lack thereof) Environmental moisture has been shown to be positively correlated to future tropical cyclone intensity trends (Emanuel 1988; Holland 1997; DeMaria and Kaplan 1999; Knaff et al. 2003b) although their affects are secondary to the effects of SST, ocean heat content and vertical wind shear. The advection of Saharan dust over the tropical Atlantic is symptomatic of an increased low-level (~700 hPa) easterly jet that propagates westward from the northwest African continent. This low-level dry-air surge can cause a marked increase in vertical wind shear and dry air entrainment that can act to influence tropical cyclones that encounter it (Dunion and Velden 2002a). Recently developed

74

multispectral Geostationary Operational Environmental Satellite (GOES) infrared imagery detects the SAL’s entrained dust and dry air as it moves westward over the tropical Atlantic (Dunion and Velden 2004). This imagery reveals that as the air masses overtake a tropical cyclone the convective intensity and organization can be reduced resulting in a general weakening and that as the tropical cyclones emerge from the air mass intensification can subsequently occur (Dunion and Velden 2002b). The effects of dry air intrusion on simulated landfalling hurricanes are investigated by Kimball (2006). Primary results are that storms with a small radial extent of moisture develop minimal rainbands and weaken as dry air from the 800–850-hPa layer wraps cyclonically and inward around the storm core. Storms with a large radial extent of moisture develop into storms with large rainbands, having smaller intensification rates initially, but then continue to intensify for a longer period of time. For these cases, the rainbands act as a barrier between the moist core and the dry environment, preventing dry air from penetrating the storm core. In the absence of land, a hurricane can sustain itself in a dry environment, provided its moist envelope is large enough. 1.1.7 Summary, forecast challenges, and recommendations for future directions Clearly significant challenges exist in forecasting tropical cyclone structure and intensity change during interaction with dynamic environments. A tropical cyclone moving into an exceptionally low shear environment may cause increased intensity forecast errors. A weakening of the tropical cyclone in response to an unexpected encounter with an enhanced vertical wind shear environment can cause over-forecasts of intensity increase. In addition, an encounter between a tropical cyclone and a midlatitude trough presents many different challenges: is it a “good trough” or a “bad trough,” and does the “good” part of a “good trough” interaction depend on the difference between the tropical cyclone’s current intensity and its MPI? It is also difficult, at times, to diagnose the current intensity and wind structure associated with tropical cyclones. In addition, current intensity forecast models are seldom able to outperform forecasts derived from climatology and persistence (Gross 1999; JTWC 2002), and only recently has a systematic way to forecast and verify wind radii information been developed (McAdie 2002). Studies involving model simulations show promise in fundamental understanding of the physical processes occurring during intensity and structural changes of tropical cyclones that occur due to environmental forcing. However, much work clearly still is needed in order to translate this understanding to useful guidance for forecasters. Studies using satellite remote sensing products to enhance general understanding of the effects of the environment on future tropical cyclone intensity and structure change also show much promise and may prove to be a fruitful way to bridge the gap between theoretical knowledge and practical guidance for forecasters. Currently there are several efforts underway to develop useful techniques for forecasting tropical cyclone intensity. Examples of these efforts include but are not limited to: Multimodel superensemble approaches to forecasting tropical cyclone intensity (Kumar et al. 2003), probabilistic forecasts of tropical cyclone intensity (Weber 2005), the addition of microwave satellite information to the SHIPS model (Jones et al. 2006), and the addition of low-level cloud-drift winds from GOES into the H-wind algorithm (Dunion et al. 2002). Techniques that have been transitioned in the last 4-5 years include but are not limited to: the operational use of AMSU-derived intensities and wind radii estimates; the implementation of STIPS to the western North Pacific (Knaff et al. 2005); the addition of ocean heat content and Geostationary IR satellite information to the SHIPS model (DeMaria et al. 2005); the H-wind algorithm (Powell, 2002); and a rapid intensification probability index for the Atlantic Basin (Kaplan and DeMaria 2002).

75

1.1.8 References Black. M. L., J. F. Gamache, F. D. Marks Jr., C. E. Samsury, and H. E. Willoughby, 2002: Eastern Pacific Hurricanes Jimena of 1991 and Olivia of 1994: The effect of vertical shear on structure and intensity. Mon. Wea. Rev., 130, 2291-2312. Bosart, L. F., C. S. Velden, W. E. Bracken, J. Molinari, and P. G. Black, 2000: Environmental influences on the rapid intensification of Hurricane Opal (1995) over the Gulf of Mexico. Mon. Wea. Rev.,128, 322-352. Brueske, K. F., and C. Velden, 2003: Satellite-based tropical cyclone intensity estimation using the NOAA-KLM series Advanced Microwave Sounding Unit (AMSU). Mon. Wea. Rev., (conditionally accepted) Corbosiero, K.L., and J. Molinari, 2002a: The effects of vertical wind shear on the distribution of convection in tropical cyclones. Mon. Wea. Rev., 130, 2110-2123. Corbosiero, K.L., and J. Molinari, 2002b: The relationship between storm motion, vertical wind shear and convective asymmetries in tropical cyclones. Accepted to J. Atmos. Sci. DeMaria, M., and J. Kaplan 1999: An updated statistical hurricane intensity prediction scheme (SHIPS) for the Atlantic and Eastern North Pacific basins. Wea. Forecasting. 14, 326-337. DeMaria, M., M. Mainelli, L. K. Shay, J. A. Knaff and J. Kaplan. 2005: Further Improvements to the Statistical Hurricane Intensity Prediction Scheme (SHIPS). Wea. Forecasting, 20, 531–543. DeMuth, J. L., M. DeMaria, J. A. Knaff, and T. H. Vonder Haar, 2003: Validation of an advanced microwave sounder unit (AMSU) tropical cyclone intensity and size algorithm. Submitted to J. App. Met. Dunion, J., and C. S. Velden, 2002a: Satellite applications for tropical wave/tropical cyclone tracking. Proceedings of the 25th Conference on Hurricanes and Tropical Meteorology, 29 April - 3 May 2002, San Diego, CA. pp 132-133. Dunion, J., and C. S. Velden, 2002b: The impact of the Saharan air layer on Atlantic tropical cyclone activity. Minutes of the 56th Interdepartmental Conference. New Orleans, LA., Office of the Federal Coordinator for Meteorological Services and Supporting Research, U.S. Department of Commerce. In Press. Dunion, J. P., S. H. Houston, C. S. Velden and M. D. Powell. 2002: Application of Surface-Adjusted GOES Low-Level Cloud-Drift Winds in the Environment of Atlantic Tropical Cyclones. Part II: Integration into Surface Wind Analyses. Mon. Wea. Rev., 130, 1347–1355. Dunion, J. P., and C. S. Velden. 2004: The Impact of the Saharan Air Layer on Atlantic Tropical Cyclone Activity. Bull. Amer. Meteor. Soc., 85, 353–365. Emanuel, K. A., 1988: The maximum intensity of hurricanes. J. Atmos. Sci., 54, 1143-1155. Emanuel, K. A., C. DesAutels, C. Holloway, and R. Korty, 2004: Environmental Control of Tropical Cyclone Intensity. J. Atmos. Sci., 61, 843–858. Frank, W. M., and E. A. Ritchie, 1999: Effects of environmental flow upon tropical cyclone structure. Mon. Wea. Rev., 127, 2044-2061. Frank, W. M., and E. A. Ritchie, 2001: Effects of vertical wind shear on hurricane intensity and structure.

76

Mon. Wea. Rev., 129, 2249-2269. Gallina, G. M., and C. S. Velden, 2002: Environmental vertical wind shear and tropical cyclone intensity change utilizing enhanced satellite derived wind information. Proceedings of the 25th Conference on Hurricanes and Tropical Meteorology, 29 April - 3 May 2002, San Diego, CA. pp172-173. Gross, J. M., 1999: 1998 National Hurricane Center forecast verification. Minutes of the 53rd Interdepartmental Conference. Biloxi, MS., Office of the Federal Coordinator for Meteorological Services and Supporting Research, U.S. Department of Commerce. pp. B24-B63. [Available from Office of the Federal Coordinator for Meteorological Services and Supporting Research, 8455 Colesville Road, Suite 1500, Silver Spring, MD, 20910]. Hanley, D. E., 1999: The effect of trough interactions on tropical cyclone intensity change. Ph. D. thesis, State University of New York at Albany, 164 pp. Hanley, D. E., 2002: The evolution of a hurricane-trough interaction from a satellite perspective. Wea. Forecasting, 17, 916-926. Hanley D. E., J. Molinari, and D. Keyser, 2001: A composite study of the interactions between tropical cyclones and upper-tropospheric troughs. Mon. Wea. Rev., 129, 2570-2584. Holland, G. J., 1997: The maximum potential intensity of tropical cyclones. J. Atmos. Sci., 54, 2519-2541. Hong, X., S. W. Chang, S. Raman, L. K. Shay, and R. Hodur, 2000: The interaction between Hurricane Opal (1995) and a warm core ring in the Gulf of Mexico. Mon. Wea. Rev., 128, 1347-1365. Jones, S. C., 2000: The evolution of vortices in vertical shear. III: Baroclinic vortices. Quart. J. Roy. Meteor. Soc., 126, 3161-3186. Jones, S. C., 2004: On the ability of dry tropical-cyclone-like vortices to withstand vertical shear J. Atmos. Sci., 61, 114-119. Jones, T. A., D. Cecil, and M. DeMaria. 2006: Passive-Microwave-Enhanced Statistical Hurricane Intensity Prediction Scheme. Wea. Forecasting, 21, 613–635. JTWC, 2002: JTWC TC Climatology Tables – North Western Pacific. [Available on-line from https://www.npmoc.navy.mil/jtwc/climostats/fcsttrkerr.html] Kaplan, J., and M. DeMaria, 2002: Estimating the probability of rapid intensification using the SHIPS model output: Some preliminary results. Proceedings of the 25th conference on Hurricanes and Tropical Meteorology. 29 April-3 May 2002, San Diego, CA. 123-125. Kimball, S. K., and J. L. Evans, 2002: Idealized simulations of hurricane-trough interaction. Mon. Wea. Rev., 130, 2210-2227. Kimball, S. K., 2006: A Modeling Study of Hurricane Landfall in a Dry Environment. Mon. Wea. Rev.,134, 1901–1918. Knaff, J. A., J. P. Kossin, and M. DeMaria, 2003a: Annular hurricanes. Wea. Forecasting, 18, 204-223. Knaff, J. A., M. DeMaria, and C. R. Sampson 2003b: Statistical, five-day tropical cyclone intensity forecasts derived from climatology and persistence. Wea. Forecasting, 18, 80–92.

77

Knaff, J. A., S. Seseske, M. DeMaria, and J. L. Demuth, 2004: On the influence of vertical wind shear on symmetric tropical cyclone structure derived from AMSU. Mon. Wea. Rev., 132, 2503-2510. Knaff, J. A., C. R. Sampson, and M. DeMaria. 2005: An Operational Statistical Typhoon Intensity Prediction Scheme for the Western North Pacific. Wea. Forecasting, 20, 688–699. Kumar,T. S. V. V., T. N. Krishnamurti, M. Fiorino and M. Nagata. 2003: Multimodel Superensemble Forecasting of Tropical Cyclones in the Pacific. Mon. Wea. Rev. 131, 574–583. Kwok, J. H. Y., and J. C. L. Chan. 2005: The Influence of Uniform Flow on Tropical Cyclone Intensity Change. J. Atmos. Sci., 62, 3193–3212. McAdie, C., 2002: Derivation of a CLIPER model for the forecast of tropical cyclone wind radii. Minutes of the 56th Interdepartmental Conference. New Orleans, LA., Office of the Federal Coordinator for Meteorological Services and Supporting Research, U.S. Department of Commerce. In Press. Möller, J. D., and L. J. Shapiro, 2002: Balanced contributions to the intensification of Hurricane Opal as diagnosed from a GFDL model forecast. Mon. Wea. Rev.,130, 1866-1881. Patra, R, 2004: Idealised modelling of tropical cyclones in vertical shear: the role of saturated ascent in the inner core. Preprints of the AMS 26th Conference on Hurricanes and Tropical Meteorology. Peng, M. S., B-F. Jeng, and R. T. Williams, 1999: A numerical study on tropical cyclone intensification. Part I: Beta effect and mean flow effect. J. Atmos. Sci., 56, 1404-1423. Persing, J., M. T. Montgomery, and R. E. Tuleya, 2002: Environmental interactions in the GFDL hurricane model for Hurricane Opal. Mon. Wea. Rev., 130, 298-317. Powell, M., 2002: An operational real-time hurricane wind analysis system (H*Wind): A JHT project for transition to operations at NHC. Minutes of the 56th Interdepartmental Conference. New Orleans, LA., Office of the Federal Coordinator for Meteorological Services and Supporting Research, U.S. Department of Commerce. In Press. Reasor, P. D., M. T. Montgomery, F. D. Marks Jr. and J. F. Gamache, 2000: Low-wavenumber structure and evolution of the hurricane inner core observed by airborne dual-Doppler radar. Mon. Wea. Rev., 128, 1653-1680. Reasor, P. D., M. T. Montgomery, and L. D. Grasso, 2004: A new look at the problem of tropical cyclones in vertical shear flow: Vortex resiliency. J. Atmos. Sci., 61, 3-22. Ritchie, E. A., 2004: Tropical Cyclones in Complex Vertical Shears. Proceedings of the 26th Conference on Hurricanes and Tropical Meteorology, 3 – 7 May 2004, Miami, FL. Ritchie, E. A., and R. L. Elsberry, 2001: Simulations of the transformation stage of the extratropical transition of tropical cyclones. Mon. Wea. Rev., 129, 1462-1480. Ritchie, E. A., and W. M. Frank, 2006a: Interactions between simulated tropical cyclones and an environment with a variable Coriolis parameter. Mon. Wea. Rev. (in press). Ritchie, E. A., and W. M. Frank, 2006b: On the interaction between vertical wind shear and tropical cyclones on a variable-f geometry: Easterly shear versus westerly shear. Mon. Wea. Rev. (in preparation).

78

Rogers, R., S. Chen, J. Tenerelli, and H. Willoughby, 2003: A Numerical Study of the Impact of Vertical Shear on the Distribution of Rainfall in Hurricane Bonnie (1998). Mon. Wea. Rev., 131, 1577–1599. Shapiro, L. J., 1983: Asymmetric boundary layer flow under a translating hurricane. J. Atmos. Sci., 40, 1984-1998. Shapiro, L. J., and J. D. Möller, 2002: Influence of atmospheric asymmetries on the intensification of Hurricane Opal: Piecewise PV inversion diagnosis of a GFDL model forecast. Extended Abstracts, 25th Conference on Hurricanes and Tropical Meteorology, Amer. Met. Soc., 251-252. Shay, L. K., G. J. Goni, and P. G. Black, 2000: Effects of a warm oceanic feature on Hurricane Opal. Mon. Wea. Rev., 128, 1366-1383. Weber, H. C., 2005: Probabilistic Prediction of Tropical Cyclones. Part II: Intensity. Mon. Wea. Rev., 133, 1853–1864. Wong, M. L. M., and J. C. L. Chan, 2004: Tropical Cyclone Intensity in Vertical Wind Shear. J. Atmos. Sci., 61, 1859–1876. Zehr, R. M., 2003: Environmental vertical wind shear with Hurricane Bertha (1996). Wea. Forecasting., In Press. Zeng, Z., Y. Wang, and C-C Wu, 2006: Environmental dynamical control of tropical cyclone intensity – An observational study. Mon. Wea. Rev., (In Press). Zhang D.-L., C. Q. Kieu, 2005: Shear-forced vertical circulations in tropical cyclones, Geophys. Res. Lett., 32, L13822, doi:10.1029/2005GL023146.

79

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES

Topic 1.2 : Tropical Cyclone Inner Core Dynamics Rapporteur: Jeff D. Kepert

Bureau of Meteorology Research Centre, GPO Box 1289K Melbourne, Australia

Email: J.Kepert@bom.gov.au Phone: +61+3+9669 4492 Fax: +61+3+9669 4660 Working Group: Michael Foley, Jeff Hawkins, Jim Kossin, David Nolan, Melinda Peng, Roger Smith, Yuqing Wang, Samuel Westrelin

1.2.1. Introduction The inner core of a tropical cyclone contains the strongest winds, and is therefore of considerable practical importance. Apart from the direct impact, it is here that the bulk of the sea-air energy flux that sustains the storm occurs, that much of the ocean response in the form of storm surge, waves and currents are generated, and that the processes that lead to intensity change take place. The considerable forecast challenge of intensity change depends on both the hard to observe and assimilate inner core processes, as well as on environmental forcing. In contrast, track forecasting, with the exception of small-scale trochoidal oscillations, depends on larger-scale processes in the environment that are easier to discern. The past few years have seen huge improvements in our ability to observe the inner core, both through in situ means, and by remote sensing. Examples include the GPS dropsonde, step frequency microwave radiometer, aircraft (now deployed by Taiwan and Canada as well as the USA), Doppler and conventional radar, passive microwave sensors, satellite sounders, rapid-scan satellite imagers, scatterometers, and portable towers and profilers, to name just a few. Some substantial theoretical advances, coupled with the ability to learn from ever-higher resolution simulation, have complemented the observational increases. These improvements have also led directly to operational advances, including an enhanced ability to monitor the eyewall replacement cycle, and the ability of operational NWP to predict intensity change and genesis, albeit to a limited degree. This report will detail these advances, and is organised as follows. In section 1.2.2 we consider the boundary layer, which is important to understanding and ameliorating storm impact, to diagnosing storm intensity, to understanding the sea-air and land-air interactions that govern the energy supply to and dissipation of the storm, and to predicting the ocean response. Section 1.2.3 describes the symmetric and asymmetric structure of the eye and eyewall, including advances in our ability to monitor changes in this area. Section 1.2.4 considers the multiple types of spiral bands observed. Implications for forecasting are summarised in section 1.2.5, a summary of conclusions presented in section 1.2.6, and recommendations for research and operations made in section 1.2.7.

80

1.2.2. Boundary Layer It is less than a decade since the first operational deployment of a GPS dropsonde (Hock and Franklin 1999) into a hurricane. This and other new observing technologies, coupled with theoretical advances, have resulted in an explosion in knowledge of the tropical cyclone boundary layer (TCBL), and the overturning of some long-established ideas and practices. 1.2.2.1 Mean structure a) Observations Observed wind profiles in tropical cyclones (TCs) frequently show a marked low level wind maximum. This maximum typically occurs around 300 to 800 m height near the eyewall, and 1 to 2 km at larger radius, and has been observed by dropsonde (e.g. Franklin et al., 2003), wind profiler (e.g. Knupp et al., 2000, 2005), and Doppler radar (e.g. Marks et al., 1999), although it has been the advent of the GPS dropsonde that has emphasised the frequent occurrence of this jet. The broad maximum is generally more-or-less obscured by smaller-scale fluctuations due presumably to turbulence, thus some form of averaging is needed to expose it clearly. For example, Fig 1.2.1 shows the observed mean normalised wind speed profile from the eyewall of seven hurricanes, in which this feature is clearly apparent. Below the jet, in the lowest 100 – 200 m, the wind speed increases nearly logarithmically with height, (Franklin et al., 2003; Powell et al., 2003), consistent with classical theory for a neutrally-stratified surface layer. A substantial amount of between-storm variability is obvious in Fig 1.2.1. The strength of the normalised maximum varies from 1.12 to 1.3, and its height from 300 to 800 m, while the strength of the normalised surface wind speed is between 0.82 and 0.96. Franklin et al. (2003) speak of the differing “character” between individual storms; we will return to this matter later. Along with variation between storms, there is substantial variation within each storm. The decrease in height of the low-level jet with decreasing radius becomes quite marked across the eyewall, and continues to the centre of the storm (Franklin et al., 2003; Kepert, 2002c, 2006a, b). Fig 1.2.2 shows the observed mean storm-relative wind profile in several annular regions in Hurricane Mitch (1998); it is clear that the depth of the frictional inflow layer and the jet height decrease towards the centre. Moreover, the azimuthal-wind maximum is generally near the top of, but within, the frictional inflow layer. Within the eye itself, both individual and mean soundings generally show little, if any, evidence of frictional retardation at the surface. This radial variation in wind structure is accompanied by a variation in the surface wind factor (SWF); that is, the ratio of the near-surface wind speed to that at some higher level. Franklin et al. (2003) showed that the widely used value of 0.8 is appropriate for the outer vortex, but that this increased to 0.9 near the eyewall. They also found that the SWF varied with reference height, recommended higher values in the outer vortex near convection than in its absence, and noted higher values on the left of the storm track than on the right. These new values revised long-standing operational practice at the NHC, and were an important influence on the reanalysis of Hurricane Andrew’s landfall intensity to Saffir-Simpson category 5 (Landsea et al., 2004). The SWF has been analysed in individual storms by Kepert (2002c, 2006a,b) and Schwendike (2004). The increase towards the storm centre is clearly marked, and some storms display higher values on the left of track than on the right. However other factors, including proximity to land in the case of Hurricane Mitch (1998), can also produce a significant asymmetry. Surface wind data from the airborne step-frequency microwave radiometer (SFMR) usually shows an increase in SWF towards the centre, and often a left-right asymmetry as well (Mark Powell, personal communication, 2006). There is also a marked azimuthal variation in TCBL wind structure. Fig 1.2.3 shows the observed profiles in Hurricane Georges (1998). It is clear that a large part of the variability between profiles is due

81

to their position within the storm (Kepert, 2006a); this is especially striking as observations nearby in storm-relative space are not necessarily nearby in time. The height and strength of the low-level jet, and the depth and strength of the inflow layer, vary consistently around the storm. Analyses of Hurricanes Mitch (Kepert, 2006b), Danielle and Isabel (Schwendike, 2005) similarly show a consistent spatial variation of wind profile shape within each storm. Several of the theoretical studies, reviewed below, have predicted that the upper boundary layer jet is supergradient. Balance in this situation has been analysed by Kepert (2002c, 2006a,b) and Schwendike (2005). They found that Hurricanes Mitch (1998) and Isabel (2002) had azimuthal-mean jets that were ~15% supergradient, but that Georges (1998) and Danielle (1998) did not. These differences highlight the inter-storm differences mentioned above, and are discussed further below. Several studies have discussed boundary layer (BL) asymmetries due to proximity to land. Here, the higher roughness over land induces increased inflow, which is typically dryer, and produces a flow asymmetry that may extend into the eyewall. Analyses of these phenomena have been presented for Hurricanes Bonnie (Schneider and Barnes, 2005), Danny (Kepert, 2002a), Floyd (Kepert, 2002b) and Mitch (Kepert 2006b). This issue is discussed in more detail in topic 0.2. Boundary-layer thermodynamics was analysed along inflow trajectories in Hurricane Bonnie (1998) by Wroe and Barnes (2003). They found little increase in inflow θe to within about 1.5 times the RMW, despite surface fluxes of over 500 W m-2, since fluxes through the top of the BL, including those due to convective cells, remove moist entropy from the BL at the same rate as the sea supplies it. Inwards of this, the storm secondary circulation suppresses convection while the surface fluxes continue to increase, giving an increase in θe. To achieve balance, they found it was necessary to either allow for some entrainment throught the BL top, or for dissipative heating. The energetics of the BL are of prime importance, in light of the considerable sensitivity of storm intensity to the energy content of the BL air beneath the eyewall.

82

Figure 1.2.1. Mean observed eyewall wind speed profile in seven hurricanes, normalised by the wind speed at 700 hPa. From Franklin et al. (2003).

Figure 1.2.2. Observed wind profiles in Hurricane Mitch (1998). (a) Mean profiles of storm-relative azimuthal wind over radius ranges 0 – 15 km (heavy) and 40 – 100 km (light). (b) As for (a), over radius ranges 15 – 25 km (heavy) and 25 – 40 km (light).(c, d) As for (a,b), but for the storm-relative radial wind component. From Kepert (2006b).

83

Fig 1.2.3: Profiles of the storm-relative azimuthal and radial wind components observed by dropsondes (curves with small-scale fluctuations) and represented in the model (smooth curves) in and near the eyewall of Hurricane Georges (1998). The model values were interpolated from the model grid to the observed dropsonde trajectory. The storm-relative position of each sonde as it fell through a height of 1 km and the storm motion are shown in the central panel. From Kepert (2006a).

84

b) Theory and Modelling A satisfactory theory of the TCBL is thus challenged to explain the structure variation between storms, and with radius and azimuth within an individual storm. Idealised models of the TCBL date back to the 1960’s and beyond (see the review in Kepert 2002c, Chapter 1). Their dimension serves to broadly classify them: 1-D column (e.g. Moss and Rosenthal, 1975; Powell, 1980), 1-D depth-averaged axisymmetric (e.g. Smith, 2003), 2-D axisymmetric (e.g. Rosenthal, 1962; Kuo, 1971, 1982; Eliassen and Lystad, 1977; Mallett, 2002; Montgomery et al. 2001), 2-D depth-averaged (Shapiro, 1983), and 3-D (Kepert, 2001; Kepert and Wang, 2001). Several of the 2-D axisymmetric models, and the 3-D models, display low-level wind maxima. These have much in common with the observations; the azimuthal maximum is contained within the frictional inflow layer, the maximum becomes more marked towards the centre of the storm, and the depth of the BL decreases towards the centre of the storm. This last property is because the BL depth varies inversely with the square root of the inertial stability (Rosenthal, 1962; Eliassen and Lystad, 1977; Kepert, 2001). These authors show that the TCBL is a modified Ekman spiral, in which the inertial stability parameter I replaces the Coriolis parameter f, and the spiral is “stretched” in the cross-stream direction by a factor ((f + 2V/r)/(f + V/r + dV/dr))1/2. Here, the upper-BL wind maximum is similar to that in the Ekman spiral, and a few percent supergradient in these simple models. These simpler models ignore the influence of vertical advection; including this process gives a markedly more supergradient wind maximum. Analysis of the momentum budget equations shows that the supergradient flow is generated by inwards advection of absolute angular momentum. The inflow is ultimately frictionally generated, but is maintained at the jet height against the outwards acceleration due to gradient imbalance by diffusive and advective transport from below; thus it is stronger in a model that contains vertical advection (Kepert and Wang 2001). Kepert (2006b) further relates this effect to the Ekman-like solution. If the vertical advection is zero or neglected, the oscillation and decay length-scales in the Ekman-like solution are equal. Introducing vertical advection makes these scales unequal; in an updraft, the oscillation scale is longer than that for decay, while in a downdraft the opposite applies. Thus the height-variation of the flow will exhibit larger oscillations near the BL top in an updraft than in a downdraft. This effect is strongest where upwards motion is strongest, so the BL jet is most marked beneath the eyewall and near rainbands. The height-resolving 2-D and 3-D models generally show an increase in the SWF towards the centre of the storm; in fact, such predictions preceded the observations of this phenomenon. The depth-averaged axisymmetric and 2-D models show a similar increase in the relative strength of the BL-mean wind towards the centre. In both cases, this is due to advection of angular momentum by the frictional inflow maintaining relatively stronger near-surface winds than in BLs with straight flow, and the effect is strongest where both the inflow and radial gradient of angular momentum are strong; that is, near the eyewall (and possibly also near strong convective updrafts). It is important to note that these models do not include any enhancement of the turbulent transport by moist convection, but that the effect is purely dynamical. Thus Franklin et al’s (2003) explanation for the reason for the observed SWF increasing towards the core and in areas of convection is likely incorrect. Having established that radial advection plays an important role in shaping the structure of the axisymmetric BL, it is perhaps not surprising to learn that azimuthal advection also has a strong influence. Kepert (2001) showed that the motion-induced asymmetry has the horizontal structure of a wavenumber-1 inertia wave. Such waves normally propagate, with the phase speed varying rapidly with radius and are not observed. However, the effect of vertical diffusion is to retard the wave propagation: the vertical tilt (in azimuth) of the phase lines and decay of the wave amplitude adjust so as to bring the wave to a halt, locked in position with the asymmetric friction forcing it at the surface. There are two such waves, corresponding to the anticyclonically- and cyclonically-propagating inertia waves, but the stalled version of the anticyclonically-propagating one dominates. It rotates

85

anticyclonically with increasing height, and has a depth scale several times that of the symmetric component. Kepert and Wang (2001) used a numerical model to relax some of the assumptions in Kepert’s (2001) analytical model, and also presented budget analyses that emphasised the role of horizontal advection in creating the spatial variation in BL wind structure. The inclusion of nonlinear processes, including vertical advection, tended to strengthen the jet, as in the axisymmetric case. This stalled wave structure, when combined with the symmetric component, is able to explain several well-known features of the TCBL. In the northern hemisphere, the surface earth-relative wind maximum is in the right forward quadrant, and the inflow angles are greatest on the right of track and least on the left. It also predicts the more recent observational findings, including that the BL jet is more marked, more strongly supergradient, and closer to the surface on the left of track, and that the SWF is higher on the left than on the right. These higher-dimension models demonstrate an important fact: that the TCBL, unlike much of the rest of the atmospheric BL, cannot be satisfactorily understood by 1-D, horizontally homogeneous (i.e., column) models. Rather, radial advection of angular momentum by the frictional inflow and asymmetric frictional forcing play a crucial role in determining the BL structure and depth. Variations in angular momentum advection play an important role in determining the spatial variation in TCBL wind profile structure. The radial gradient in angular momentum varies greatly between storms – some storms have relatively “flat” radial variation in the wind strength, or equivalently, are inertially stable and have a radial gradient of angular momentum throughout, while others have “peaked” profiles, and are inertially near-neutral with weak angular momentum gradient outside the RMW. Kepert and Wang (2001) presented model calculations of the resulting BL structure for these extremes. The “flat” case had a weakly (5 – 10%) supergradient jet extending from the RMW to large radii, while the “peaked” one had a strongly (25%) supergradient jet confined to the vicinity of the RMW. The frictional inflow and eyewall updraft are also relatively stronger in the “peaked” case. Observational confirmation of these differences has been provided by Kepert (2006a,b) and Schwendike (2005), who found that of four storms analysed (Danielle, Georges and Mitch of 1998, and Isabel of 2002), two had markedly supergradient flow in the upper BL beneath the eyewall, and two were indistinguishable from balance. The difference in the structure of these storms is as predicted by Kepert and Wang (2001): the “peaked” storms had supergradient flow, while the “flat” ones didn’t. Fig 1.2.3 contains the modelled wind profiles in Georges corresponding to the dropsonde observations, and shows that the model is able to reproduce much of the around-storm variation in structure, in both the azimuthal and radial flow components. Thus these analyses provide strong confirmation of the theoretical predictions, being able to explain not just the general features, but also the differences between storms. Further evidence of the ability of these models to predict the differences between storms was provided by Kepert’s (2004) preliminary analysis of the eyewall wind profiles discussed by Franklin et al. (2003). The Kepert and Wang (2001) model, forced by aircraft observations of radial storm structure, and combined with a crude calculation of the baroclinic warm core effect, was able to largely reproduce these differences. Thus the difference in “character” of eyewall wind profile between storms noted by Franklin et al. (2003), is seen to have a dynamical cause. c) Stability Effects Stability is known to have a profound influence on the atmospheric BL. Richardson number-based arguments show this effect will be smaller in the TCBL than elsewhere, but not absent. Powell and Black (1990) demonstrate that it is indeed important, producing a variation in the observed SWF of ~0.1, of similar magnitude to the dynamical variations discussed above. Little further work has been done on this factor, but it is clearly important to a complete understanding, and the ability to accurately predict, the TCBL flow. d) Operational Implications Theoretical predictions of spatial variation in the SWF, supported by observational analyses, have led

86

to changes in the way aircraft-observed winds are used to estimate the surface wind. This has resulted in an increase in the estimated intensity of storms with aircraft reconnaissance, and to the reanalysis of historical storms to higher intensity, such as Hurricanes Andrew (Landsea et al., 2004) and Donna (Dunion et al., 2003). This increased knowledge will also impact in storm analysis when aircraft reconnaissance is not available, since analysis of such storms rests heavily on the Dvorak technique (see Velden et al. 2006 for a review), which is calibrated against aircraft data. There are also implications for areas such as central pressure – maximum wind relationships, and the formulation of parametric wind models used for forcing storm surge and wave models, and for engineering and insurance risk analyses. One missing factor in such applications is the effect of eyewall slope. The eyewall is approximately an angular momentum surface, thus the maximum gradient wind will vary little with height along a near-vertical eyewall, but will increase towards the surface when the eyewall is sloped away from the vertical. Little information is available about the impact of this, although Dunion et al. (2003) present a statistical scheme that includes this effect. Kepert (2004) shows how to calculate the slope of the angular momentum surfaces from flight-level wind and temperature data under certain assumptions and uses this to calculate the surface pressure field. The derived surface pressure field (and storm motion) may then be used to force the Kepert and Wang (2001) model to account for the effect of friction. 1.2.2.2 Transients and Instabilities The TCBL supports a number of instabilities and transient structures. There is some overlap between this topic and that of general instabilities leading to spiral band structures. This section considers phenomena that are clearly part of the BL, chiefly rolls. There have also been several reported cases of fine-scale spiral bands reported in the TCBL, which are larger in scale than the above cases and so it is not clear that they are the same phenomenon. Examples include Gall et al. (1998), Kusunoki and Mashiko (2006), Kusunoki (2006). The reader should refer also to the discussion in section 1.2.4.2 for fine-scale spiral bands that occur partly within the BL. a) Boundary Layer Rolls - Observations Boundary-layer rolls are very common in the atmospheric BL (see e.g. Etling and Brown 1993 for a review). Wind circulations associated with these rolls can produce highly organized and damaging surface winds (Wakimoto and Black 1994). Wurman and Winslow (1998) presented the first Doppler radar evidence for their existence in tropical cyclones, indicating intense horizontal roll vortices with an average wavelength of 600 m roughly aligned with the mean azimuthal wind in Hurricane Fran near landfall. The associated variation of wind speed was large: bands of 40 – 60 m s-1 flow alternated with 15 – 35 m s-1. Several papers have since presented similar evidence. Katsaros et al. (2002) examined SAR images of Hurricanes Mitch and Floyd and also found periodic kilometre-scale variation. More recently, Morrison et al. (2005) describe features that are significantly less streaky in appearance, to the extent that it is not entirely clear that they are the same phenomenon. The different radar technology used by the groups may have contributed to this difference. Lorsolo et al. (2006) analyse the vertical structure of the rolls, find them to be coherent through the depth of the BL (~500 m), and compare radar- and tower-measured winds with good agreement, demonstrating that the roll circulation extends to the surface, albeit with other scales of motion superimposed. b) Boundary Layer Rolls – Theory Theoretical analyses of roll development in tropical cyclones were provided by Foster (2005) and Nolan (2005). Foster (2005) argues that the tropical cyclone BL is an ideal environment for roll development. His argument extends the classical theory of roll development as an inflection-point

87

instability of the frictionally-induced cross-isobar flow to the case of a tropical cyclone. Here, the cross-stream shear and hence instability are strong because the BL is relatively shallow, and the cross-stream component in analytical solutions is stronger than in classical Ekman-like solutions for straight flow (Kepert 2001). Nolan (2005) presents a stability analysis of a symmetric vortex, and finds both symmetric and asymmetric responses. The instabilities acquire some energy from the shear in the radial flow near the top of the BL, in which regard they are similar to Foster’s (2005) rolls. However, Nolan shows that the vertical shear of the azimuthal wind can also contribute energy to the instability, and that the relative importance of these mechanisms depends on the inertial stability of the storm and on the orientation of the mode. c) Turbulence and Gusts The GPS dropsonde has now been operational for close to a decade, and several thousands have been deployed in the eyewall of hurricanes. Extreme gusts have been reported in both horizontal and vertical wind components (Aberson and Stern 2006, Henning 2006, Stern and Aberson 2006). While these are (by definition) rare events, the steadily increasing sample is beginning to enable statistical characterization of their nature. Specially-deployed towers have been used in landfalling hurricanes for several years, and are yielding valuable data on the turbulence structure. Schroeder and Smith (2003) calculated a range of turbulence statistics from data taken during the landfall of Hurricane Bonnie, and found general agreement with established gust factor curves, but additional energy in the power spectrum at low frequencies. Wind averaging periods continue to differ between operational centres, with conversion and interpretation presenting problems. Problems include with the calibration of the Dvorak technique, usually taken to provide a 1-minute mean wind speed, for use in centres which employ 10-minute averaging. d) Other small-scale features in the boundary layer Analyses of dropsonde wind data in Hurricane Georges (1998), shown in Fig 1.2.4, reveal a distinct wavenumber 3 asymmetry in the flow near the eyewall below 1 km (Kepert 2006a), which was apparently not propagating and persisted for the period of the observations (6 hours). The lack of propagation and shallow depth rule out all known TC instabilities as a cause – in particular, these are distinct to the usual eyewall mesovortices discussed in section 1.2.3.2. A similar feature, except at wavenumber 2, was seen in the BL beneath the eyewall of Hurricane Humberto (2001) (S. Feuer, personal communication, 2003). e) Operational Implications Observations and theory show that BL rolls are common in TCs, and are associated with streaks of strong winds at the surface. The presence of these coherent structures implies that wind damage will be different to that expected in a “more random” turbulent field. They may also result in more efficient transport of heat, moisture and momentum than is presently assumed in numerical models of TCs.

88

Fig 1.2.4. Analyses of the storm-relative azimuthal (upper 6 panels) and radial (lower 6 panels) wind component at several levels in Hurricane Georges (1998). The contour interval is 5 m s-1 with heavy labelled contours at multiples of 20 m s-1. Darker shading corresponds to stronger winds (upper panels) and stronger inflow (lower panels). The white circle in the lower panels indicates the approximate RMW. From Kepert (2006a).

89

1.2.3. The eye and eyewall Tropical cyclone (TC) intensity change is governed by a number of known factors. The climatology of intensity change was documented by Dvorak (1984) and showed that an average tropical storm intensifies at a rate of a few m/s per day, and an average hurricane intensifies at a rate of about 12 – 13 m/s per day. This mean intensity change typically continues for 3 – 5 days after the storm reaches tropical storm strength (18 m/s). After maximum intensity, weakening typically occurs at a slower rate. This was recently corroborated and expanded by Emanuel (2000) who showed that a storm that does not encounter land or decreasing sea surface temperatures, intensifies, on average, at a rate of about 12 m/s per day for about 5 days, and then begins to weaken at a slower rate of about 8 m/s per day. In addition to the climatology of intensity change, we also know that environmental conditions play a key role – for example, if a storm moves over colder water or land, or if the ambient environmental vertical wind shear increases, weakening typically follows. Alternatively, an environment that is not conducive for intensification can become more favorable over time, and strengthening would typically occur. Ideally then, we would be able to explain the variance from climatology of hurricane intensity change in terms of the variance of the synoptic-scale storm environment. This is not the case however, and it is fairly typical for storms to strengthen or weaken, sometimes rapidly, without any commensurate changes in the external storm environment. Although the specific processes involved remain an open question, this behaviour is widely believed to result from internal vortex-scale processes that can have a profound effect on how storm intensity evolves, and this means that our ability to model and ultimately predict hurricane intensity change is dependent on our ability to simultaneously model a very broad range of spatial scales. The vortex-scale processes to be considered include PV redistribution resulting from dynamic instability of the eyewall and PV transport by mesovortices, eyewall replacement cycles, and mean-flow amplification via PV. 1.2.3.1 Symmetric structure a) Eyewall Contraction, Expansion and Replacement The most remarkable eyewall process related to TC intensity change is the concentric eyewall and eyewall replacement processes (Willoughby et al., 1982). In this concentric eyewall model, as a TC and its primary eyewall intensifies, convection outside the primary eyewall becomes organized into a ring (the secondary eyewall) that encircles the inner eyewall and coincides with a local tangential wind maximum. As the second eyewall propagates inward and amplifies, the inner eyewall weakens and is eventually replaced by the outer eyewall. This eyewall replacement cycle is usually accompanied by a weakening and then a re-intensification of the TC; and thus producing a large fluctuation in TC intensity. While this process is well known, recent years have seen a substantial increase in our ability to monitor it, which we now review. Tropical cyclone (TC) eyewalls have been under sampled by visible/IR imagery due to persistent upper-level cloud obscuration. However, satellite microwave sensors can “see through” non-raining clouds (Spencer et al., 1989) and thus have unraveled several TC inner-core secrets by providing “snapshots” depicting eyewall evolutions (Hawkins et al., 2006). In addition, two new microwave imagers (SSMIS and WindSat) since IWTC-V have augmented the temporal sampling from these polar orbiter-based sensors and enhanced our knowledge of eyewall characteristics. The ability to “see through” non-raining clouds permits satellite analysts to more accurately monitor inner-storm structure that perplexes vis/IR imagery such as: a) early eyewall formation, embedded eyes, middle and upper-level shear and concentric eyewalls (Hawkins et al., 2001). TC multiple eyewall characteristics have been outlined in a growing number of papers as noted by Hawkins et al. (2006), McNoldy (2004), Kodama and Yamada (2005), Hawkins and Helveston (2004), an excellent

90

summary by Simpson et al. (2003), Lee et al. (2002), and Velden and Hawkins (2002). Each effort utilizes microwave data sets to view and infer eyewall configurations. Currently, a constellation of “operational” and research passive microwave imagers and sounders provide sufficient temporal coverage to capture many eyewall structure changes. The four operational imagers include the Special Sensor Microwave Imager (SSM/I, Hollinger, 1989) and the Special Sensor Microwave Imager Sounder (SSMIS, Wessel et al., 2003). A wealth of research instruments greatly aids these sensors: the TRMM Microwave Imager (TMI), the Advanced Microwave Scanning Radiometer (AMSR-E), and the Coriolis WindSat polarimetric radiometer. This suite of seven (7) microwave imagers will not likely be repeated for decades based on future global launches (National Research Council, 2004, 2006). Coarser resolution microwave sounders also help map eyewall/rainband features: the Advanced Microwave Sounding Unit (AMSU-3) and the Microwave Humidity Sounder (MHS-1). These four operational cross-track scanners are best at nadir and can assist with storms having normal or large eyewalls. TC eyewalls form as the result of rainbands spiralling into the system centre and eventually organize into a continuous circular-like band of enhanced convection. The beginning eye diameter can have a huge range in values, but eyewall diameter typically decreases as the storm intensifies. Eyewall contraction continues and reaches a “minimum” diameter at peak strength. Many storms begin formation of a secondary eyewall at a larger radius once the inner eye reaches a critical diameter. This formation process most frequently occurs at 120 kts or higher. The secondary eyewall completely encircles the inner eye, moisture and momentum flux to the inner eye declines and it decays, eventually leaving only the outer eye at a much larger radius than the first eyewall. The secondary eyewall (now main eyewall) can contract inward and begin the process again (eyewall replacement cycle-ERC). ERCs can continue for two or three times depending on environmental characteristics and are most likely to occur within western Pacific typhoons that form in the deep tropics near Guam and don’t encounter strong shear or cold SSTs (Hawkins et al., 2006). The eyewall replacement cycle was captured for Hurricane Wilma as illustrated in storm-centred 85 GHz imagery in Fig 1.2.5 (Hawkins et al., 2006). During a one week timeframe, the storm formed a single eyewall (Oct 20th) that transitioned to two eyewalls by Oct 21st as a major rainband spiralled in from the SW sector, and the small inner eyewall decayed by the 23rd as the outer eyewall become the remaining eyewall feature. The new eyewall retained its large diameter as the storm raced across south Florida, largely explaining the extended area of high damaging winds. If favourable conditions persist (warm SSTs, weak shear, no trough interactions, etc) the process can repeat, but it was interrupted by landfall and a strong trough in Wilma’s case.

91

Studies of nine years of passive microwave data reveal that 40-80% of TCs reaching 120 kts or higher obtain double or concentric eyewalls (Hawkins et al, 2006). The southern hemisphere has the lowest percentage (40%), eastern Pacific (50%), Atlantic (70%) and western Pacific (80%). These values fluctuate markedly from year to year and basin to basin, but more intense storms (Cat 3 and higher) have the greatest chance to form and maintain double eyewalls. The duration of double eyewall configuration can range from less than 12 hours to 2-3 days and is directly dependent on environmental conditions. For example, shear can stop double eyewalls in short order. A few intense double eyewall storms do not follow the typical eyewall replacement scenario, but instead form a wider single eyewall with extremely cold cloud tops (IR) and brightness temperatures (Tb- microwave). In addition, rainbands tend to dissipate or become significantly shortened as the TC takes on an “annular” outline (Knaff et al., 2003). These annular storms exist in specific shear and SST regimes, and can maintain their intensity for multiple days with maximum sustained winds in excess of 120 kts. Efforts are currently underway on automated methods to help predict which TCs will evolve into annular and concentric eyewall systems since they represent key forecast busts as the MSLP for annular storms are typically under forecast while concentric TCs are over forecast (Kossin et al., 2006; Cram et al., 2006). The initiation of concentric eyewalls has been studied in idealised models by Nong and Emanuel (2003), who show that the wind-induced surface heat exchange (WISHE) instability is necessary for their growth. They argue that a finite-amplitude, externally forced perturbation is necessary to trigger the instability, and that the growth is sensitive to the boundary-layer moisture. This will make prediction challenging, as this parameter is often poorly observed over the tropical oceans. Recently, Cangialosi

Figure 1.2.5: Time series of 85 GHz storm centered SSM/I imagery for Hurricane Wilma from Oct. 17-24, 2005. Courtesy of the Naval Research Laboratory, Monterey, CA.

92

et al. (2006) and Ortt and Chen (2006) reported success in a couple of cases of outer eyewall formation prediction, using high-resolution NWP in research mode. b) Inner-core buoyancy Recent numerical studies (Zhang et al. 2000, Braun 2002) and aircraft flight-level data analysis (Eastin 2002, 2003) show different results regarding the relative role of buoyancy and perturbation pressure gradient forces near the core and eyewall of tropical cyclones (TCs). In a numerical simulation of Hurricane Andrew (1992), Zhang et al. concluded that air in the eyewall was negatively buoyant and was forced upward by perturbation pressure gradient forces. However, in a similar simulation of Hurricane Bob (1991), Braun found that eyewall updrafts are positively buoyant with respect to an environment that includes the vortex-scale warm-core structure. Eastin examined the buoyancy of eyewall convective updrafts in hurricanes based on aircrafts flight-level reconnaissance data and found that eyewall updraft cores were positively buoyant relative to a background mesoscale environment. Smith et al. (2005) established a generalized buoyancy force with the basic state as a function of height and radial distance. The buoyancy can be distinguished as one part related to the symmetric balanced vortex (system buoyancy) and the other part associated with cloud dynamics (local buoyancy). They suggested that the discrepancy of the foregoing studies is mainly a result of different definition of the buoyancy, in particular, its reference states. In Zhang et al., a reference state that varies in time and space was obtained by performing a running mean of the numerical model output over four neighbouring grid points on a constant σ-surface. In contrast, Braun used Fourier decomposition into different azimuthal wavenumbers and selected wavenumbers 0 and 1 to define the reference state. These two representations are more akin to the local buoyancy defined by Smith et al. (2005). Eastin defined the reference state by applying a running low-pass (Bartlett) filter to data along the flight track. This method corresponds more with the definition of the local buoyancy. Although the buoyancy force may be different when derived from different reference states, the sum of the buoyancy force and the perturbation pressure gradient is unique, and it is the sum that determines the vertical motion. In addition, results from both Braun and Eastin indicate that although the updrafts associated with positive buoyancy covers only a small portion of the eyewall area, they account for a majority the upward mass flux, consistent with the concept of “hot towers” (Hendricks. et al. 2004). Zhu and Smith (2002) explained the mechanism of shallow convection in stabilizing the core region of a TC. The shallow convection transports air with low moist static energy from the lower troposphere to the boundary layer, stabilizing the atmosphere not only to itself, but also to deep convection and reduces the rate of heating and drying in the troposphere. Also it moistens and cools the lower troposphere. This reduced heating, together with the direct cooling of the lower troposphere by shallow convection, diminishes the buoyancy in the vortex core and thereby the vortex intensification rate. Regarding the structure of the buoyancy, Eastin et al. (2005a,b) further examined the buoyancy distribution in the inner core of Hurricanes Guillermo (1997) and Georges (1998) using airborne radar, dropwindsonde, and flight-level observations. Their results indicate that the low-level eye can be an important source region for buoyant eyewall convection. It was observed that buoyant eyewall updraft cores and transient convective-scale reflectivity cells are predominantly downshear and left-of-shear. Most eyewall downdraft cores that transport significant mass downward are located upshear. Negative buoyancy was most common in left-of-shear downdrafts, with positive buoyancy dominant in upshear downdrafts. Some buoyant updraft cores were encountered in the midlevel eyewall exhibit equivalent potential temperatures much higher than in the low-level eyewall, but equivalent to the observed in the low-level eye. Asymmetric low-wavenumber circulations appear to be responsible for exporting the high-θe eye air into the relatively low-θe eyewall and generating the locally buoyant updraft cores. It was suggested that an ensemble of asymmetric buoyant convection could contribute to hurricane evolution possibly with three different mechanisms. While the “hot towers” concept is popular recently in explaining the development of TC through a two-stage process (Hendricks et al. 2004), it appears that more studies may be needed to understand

93

the asymmetric buoyancy distributions during a decaying as well as an intensifying stage to understand the thermodynamics associated with. c) Parametric Representation Many practical applications require a simple representation of the surface wind field of a TC, dependent on only a few parameters. Examples include storm surge modelling, risk analysis, and engineering design. Of the several extant, the parametric profile of Holland (1980) has been most widely used for such applications, usually with an ad hoc representation of the motion asymmetry and boundary layer friction added. The Holland profile has several deficiencies, including that the wind maximum is insufficiently sharp, and that the decay of wind speed at large radius may be incorrect (Willoughby et al., 2004). A new family of profiles (Willoughby et al. 2006) overcomes these problems, although at the cost of having more free parameters, and being less tractable mathematically. Mallen et al. (2005) studied the radial profile of vorticity using aircraft data and showed that some parametric profiles are deficient, in that they may have an annulus of negative relative vorticity. Such an annulus is not present in observations, while Reasor et al. (2004) discuss theoretical reasons for its absence being important for the ability of the vortex to resist shear. Although Mallen et al. did not consider the Holland (1980) and Willoughby et al. (2006) profiles, the former always has this problem, while the latter may if parameter values are not carefully chosen. Many applications require near-surface winds, while the profiles deliver a gradient-level wind. Crude ad hoc parameterisations of friction and the motion-induced asymmetry are customarily used. The analytical boundary-layer model of Kepert (2001) would seem to offer advantages here, since it includes much of the relevant physics, but has so far not been widely adopted for this purpose. d) Radar Observations To be relevant, observations of the inner core region of a tropical cyclone need to be available at high temporal and spatial resolution. These requirements can be satisfied with radar. Doppler radar observations (ground-based or airborne) stand as powerful tools to capture the three-dimensional wind structure and reflectivity field. But the Doppler velocities can not be used immediately and need to be processed by some sophisticated algorithms to become interpretable. The VTD (velocity track display) method can retrieve the symmetric part of the tangential and radial wind (Lee et al., 1999; Lee and Marks 2000; Lee et al., 2000) and EVTD (extended velocity track display) method can determine wavenumber-0 and -1 components of tangential and radial wind; this last one was also extended to ground-based radar by Roux et al, 2004 (GB-EVTD: ground-based extended velocity track display). Recently, Liou et al. (2006) used two radars in a very similar method. The GB-EVTD technique has been applied to the intense cyclone Dina (Roux et al, 2004) when its elliptical cyclonically rotating eye passed closest (<130 km) to La Réunion island. The three-dimensional dynamical structure of the inner core region and structure of the rainbands is well seen and very instructive. The rotating maximum tangential wind is correlated to the presence of maximum reflectivity values. The steep orography of La Réunion affects the basic flow and the cyclone wind field as well which induces modifications in the internal structure of the cyclone. The maximum tangential wind migrates from east to south which can locally bring to very variable impact on such a small island; this information has therefore a great impact on the nowcasting. Nuissier et al (2005) used airborne Doppler radar data to define a specified vortex for model initialization. They removed the ill-defined, too weak and misplaced vortex in global analyses (Kurihara et al., 1993, 1995) and replaced it by a balanced vortex deduced from airborne Doppler radar data.

94

1.2.3.2 Asymmetries of and inside the eyewall a) Eyewall mesovortices The existence of small vortices, of ~10km scale, abutting the inner edge of the eyewall, has been known for some time, and is associated with the occasional occurrence of a markedly polygonal structure of the inner edge of the eyewall on radar reflectivity images (e.g. Muramutsu 1986). These polygonal eyewalls had typically between 3 and 6 sides, and rotated somewhat more slowly than the azimuthal flow. Aircraft observations (Marks and Black, 1990; Black and Marks, 1991) showed that in extreme cases, an eyewall mesovortex (EMV) could be associated with local wind and pressure perturbations of magnitude approaching that of the primary vortex core. Recently, new observations and improved physical understanding has shown that these EMVs are quite common, represent an additional hazard to humanity, and play a role in the overall dynamics of the storm. A spectacular recent example was the six EMVs observed in Hurricane Isabel, shown in Fig 1.2.6 (Kossin and Schubert, 2005). Kossin and Eastin (2001) used aircraft data to show that the wind and thermodynamic structure of strong TCs evolve between two distinct regimes. In regime I, the radial profile of wind across the eye is U-shaped, with maximum angular velocity within the eyewall, and an annular ring of vorticity just inside of the RMW. The eye is typically warm and dry, with elevated values of θe in the eyewall and lower values within the eye. Regime II, in contrast, is characterised by a V-shaped wind structure within the eye, with vorticity and angular velocity maxima near the centre of the eye. The air in the eye is relatively moist, with a θe maximum near the vortex centre. Transitions from regime I to II can be very rapid, occurring in less than an hour. The simultaneous change in kinematic and thermodynamic variables suggests the sudden onset of intense horizontal mixing between the eyewall and the eye. This mixing appears to be caused by the onset of eyewall mesovortices (EMVs), consistent with the barotropic instability of regime I. Kossin and Schubert (2003) discuss how this mixing is distinct from a horizontal diffusion process. This barotropic instability is similar to that in other strongly sheared flows. The regime I hurricane is approximately a hollow tower of vorticity (located just inside of the RMW), with less vortical flow inside and out. Vortex Rossby waves can propagate on both high vorticity-gradient surfaces of the tower: on the inner face, the propagation is with (i.e. faster than) the flow, while on the outer face, it is against (slower). It is thus possible that the inner and outer waves can phase lock and mutually amplify, leading to exponential growth of the waves. This process was modelled in an unforced barotropic model by Schubert et al. (1999), who showed that the vortex ring could break down into several mesovortices, which subsequently merge to form a vortex monopole. Thus the wind profile went from U-shaped to V-shaped; precisely the change later noted in aircraft data by Kossin and Eastin (2001). A range of final vortex structures is possible, depending on the size, width and strength of the initial vortex ring. Kossin and Schubert (2001) explore the part of the parameter space relevant to TCs. The typical response is for the vortex ring to break into a relatively large number of mesoscale vortices, which subsequently merge. These mergers may proceed to give a single discrete vortex monopole, or to a quasi-stable “vortex crystal”, an asymmetric lattice of mesovortices rotating as a solid body. The flow associated with such a structure consists of near-straight line segments, making up a persistent polygonal shape. Such lattices can also undergo internal rearrangements – e.g. from a pentagon with a central mesovortex to a hexagon, and back again. The former of these structures is strikingly similar to the well-known instance of mesovortices in Hurricane Isabel (Fig 1.2.6). These dramatic rearrangements of vorticity are accompanied (in the model) by equally spectacular pressure changes. In the change from the U-shaped (vorticity hollow tower) to the V-shaped (vortex monopole) structures, the winds inside the eye accelerate dramatically, while the maximum wind decreases. Integrating the gradient wind equation inwards (or solving a nonlinear balance equation),

95

the net effect is that the central pressure falls substantially, even while the intensity as measured by the strongest winds decreases. Similar vortices have been produced in the laboratory. Montgomery et al. (2002) describe a water-flow apparatus which produces a curved shear layer, with primary and secondary circulations and aspect ratio similar to a hurricane. Two quasi-steady vortices, together with intermittent secondary vortices, form from shear instability of the curvilinear shear layer on the inner side of the “eyewall”. The peak tangential velocity occurred within the mesovortices, and was ~50% stronger than that of the parent vortex. An eye structure, intermediate between the Kossin-Eastin regimes, associated with TCs with an unusually high degree of axisymmetry, was identified by Knaff et al. (2003) and named “annular hurricanes”, also known as “truck tyres” from their appearance on IR satellite imagery. These storms apparently form from the asymmetric mixing of eye and eyewall, possibly by mesovortices, but the mixing does not proceed all the way to the monopole structure of regime II. Annular hurricanes form within certain specific and relatively rare environmental conditions, including weak SE’ly environmental shear (in the Northern Hemisphere) and favourable thermodynamics, as measured by the potential intensity (PI). As well as the large and symmetric eye, they are also unusually symmetric outside the eye, with little evidence of outer rainbands. Significantly for forecasting, they maintain intensity longer and weaken more slowly than other TCs, and are thus a significant source of intensity forecast error. Persing and Montgomery (2003) used an axisymmetric model to test the PI theory of Emanuel (1987, 1995). They found that, when run at high resolution, the model predicted intensity averaging about 20 m/s higher than PI, which they dubbed “superintensity”. The reason for this was that the model developed small-scale vortices, aligned with the mean flow, along the inner edge of the eyewall. These vortices efficiently mix the high-θe air from the eye boundary layer into the eyewall updraft, increasing the energy content of this air and hence the storm intensity. Although their model is axisymmetric and cannot generate EMVs, they argue that EMVs might similarly transfer high-θe air into the eyewall, a process that is absent from the derivation of the PI theory. In this context, Braun (2002) presented evidence of episodic mixing of eye air into the eyewall by EMV-like features, and also showed that θe was not constant along parcel trajectories in the eyewall, due to this mixing process. Much of the interest in EMVs has been in highly symmetric storms with a clear eye in the cirrus overcast, since this facilitates their identification from satellite imagery. Recently, two studies (Braun et al., 2006; Halverson et al., 2006) have shown that they also exist in sheared storms, where the axis is tilted. A tilted axis is associated with enhanced low-level convergence and ascent on the downtilt side, leading to increased rainfall downtilt left (in the Northern Hemisphere). The opposite applies uptilt. These papers showed that EMV-like features may orbit the tilted eyewall, with their updraft and vorticity intensifying as they moved into the favourable downtilt area, and weakening as they leave it. Cyclonic advection of the enhanced convection in the EMVs lead, in the case of Hurricane Erin, to the coldest cloud tops being on the upshear side of the storm.

96

Figure 1.2.6: Defense Meteorological Satellite Program visible image of Hurricane Isabel at 1315 UTC 12 Sep 2003. The six mesovortices - one at the centre and five surrounding it - cause the starfish pattern in the eye.

b) Partial eyewalls Another eyewall process is the so-called partial eyewall replacement proposed and studied numerically by Wang (2002b). He argued that strong perturbation from an outer spiral rainband could amplify the vortex Rossby waves in the eyewall, causing a large distortion of the eyewall and partial eyewall breakdown accompanied by a weakening of the TC. The eyewall can later recover from breakdown through axisymmetrization, resulting in a re-intensification of the storm. Therefore the eyewall breakdown/recovery is accompanied by a weakening/intensifying cycle of the TC. This partial eyewall cycle is an asymmetric process, and hence is different from the symmetric eyewall replacement studied by Willoughby et al. (1982). The latter is possible only for highly symmetric, intense storms, while the former can happen to both symmetric and asymmetric, and both intense and weak, storms. 1.2.3.3 Eye Tilt There are quite different processes that can cause the eye of a tropical cyclone to tilt. The vertical shear of environmental flow in which the tropical cyclone is embedded is frequently cited as the major factor (e.g., Jones, 1995; Wang and Holland, 1996c; Frank and Ritchie, 1999; see also subtopic 1.1 report). Other processes include the so-called beta-effect (e.g. Wang and Holland 1996a, b), and interaction with another tropical cyclone or synoptic/mesoscale systems (e.g. Wang and Holland, 1995; Holland and Lander, 1993). The tilt of the inner core structure has many consequences. It complicates the interaction between the tropical cyclone and its large-scale environment because the tilted vortex involves self interaction between the upper and lower levels. Understanding the dynamics of a tilted vortex is thus as equally important as the interaction with large-scale environment. Since the atmosphere on the rotating earth is vertically stratified, any perturbation, such as a potential vorticity (PV) anomaly, on one layer can have an impact on the other through the so-called vertical penetration (Hoskins et al. 1985) or vertical coupling. As a result, a self interaction of a tilted vortex may be expected and the extent to which the interaction occurs depends strong on a so-called vertical penetration depth, which is a function of the horizontal scale of the vortex and the background

97

stratification and local rotation (Shapiro and Montgomery 1993). The first-order interaction of the tilted tropical-cyclone-like vortex is the mutual cyclonic rotation of the upper and lower vortex centres due to the vertical penetration flow associated with the inner-core PV anomalies. Given an initially tilted cyclonic vortex without any environmental flow on an f-plane, the surface vortex centre would experience a cyclonic looping motion (Smith et al., 2000; Reasor and Montgomery, 2001). Even with constant vertical shear forcing, the track wobbles or oscillations can be expected (Jones, 1995; Wang et al., 2004). However, for more realistic tropical-cyclone-like vortices that have an anticyclonic circulation in the upper troposphere, the interaction between the anticyclone aloft and cyclonic vortex below could cause a deflection of the surface track toward the left of the vertical shear vector (Wu and Emanuel, 1993; Wang and Holland, 1996c). A fundamental question is how a TC vortex can sustain a coherent vertical structure in vertical shear. Wang and Holland (1996c) found in a numerical study that the cyclonic portion of the TC could remain upright in a moderate vertical shear. The TC core undergoes successive downshear tilting during the first 24-h while realigning over a 72-h period. A quasi-steady tilt to the downshear left was found even in the case where the diabatic heating was not considered. To understand the vertical alignment of a tilted TC vortex, Reasor and Montgomery (2001) developed a new theory, which separates the mean vortex evolution from the evolution of the tilt asymmetry. From this new perspective, the vertically-averaged azimuthal mean component of a tilted vortex is defined as the mean, while the departure from this mean as the tilt perturbation. The subsequent evolution of the tilt perturbation was captured by a linear, dry vortex Rossby wave (VRW) mechanism. They found that the continuum modes in the dynamical system destructively interact with the VRW, leading to the decay of the VRW and hence the vortex tilt. Schecter et al. (2002) viewed the vertical alignment as a result of the damping of the VRW due to its interaction with the mean vortex circulation. Reasor et al. (2004) show further that the VRW damping mechanism provides a direct means of eradicating the tilt of intense TC-like vortices in unidirectional vertical shear, and that intense TC-like vortices are much more resilient to vertical shear than previously believed. Therefore, the VRW damping mechanism intrinsic to the dry adiabatic dynamics of the TC vortex may play a crucial role in maintaining the vertically coherent structure of TCs in moderate vertical shear. However, how the diabatic heating including both the symmetric and asymmetric components affects this dry dynamics is still unknown and serves as an important topic of future study. Wong and Chan (2004) showed that the secondary circulation and the associated diabatic heating reduce the vertical tilt and the weakening of the TC, but the precise mechanisms remains to be addressed. 1.2.4. Spiral Bands Along with the eye of the storm, spiral bands are one of the most recognized features of tropical cyclones. While the largest, “outer” bands have been easily seen on low-resolution satellite images for many years, increases in observational technology have led to the identification of spiral bands and “band-like” features on increasingly smaller scales, down to just a few hundred meters in size. These very small “streaks” are almost certainly a feature of the tropical cyclone boundary layer alone and are discussed above in section 1.2.2.2a; here, we will focus on bands with widths and separations distances of 4 km and larger, that extend through and above the depth of the boundary layer. 1.2.4.1 Ongoing studies of Inner-core bands While not easily seen in satellite images, inner-core bands are clearly and almost universally observed in radar images of tropical cyclones ranging from strong tropical storm to category-5 intensity. These bands are typically 10-40 km in width, separated from each other on similar and larger scales, and show distinct spiral patterns radiating outward from the inner core (although they are not always “attached” to the eyewall).

98

Considerable effort from many different investigators has been put into establishing a dynamical mechanism for these bands. The most widely evaluated mechanism is the vortex-Rossby wave theory of Montgomery and Kallenbach (1997), which postulates that the bands of precipitation are associated with waves of PV which propagate radially and vertically on the PV gradient of the symmetric vortex. For low-wavenumber (n=1 to n=2) bands, substantial support for this theory has been presented since the last IWTC report. Careful analysis of high-resolution (3-6 km) numerical simulations of both real-case and idealized storms have shown a high spatial correlation between precipitation, clouds, and PV in the bands, strongly suggesting that the bands are coupled to vortex-Rossby waves (Chen and Yau 2001; Wang 2002a, b). These results were further supported by the “empirical normal mode” analysis of Chen et al. (2003), who found that 70-80% of the wave activity of the n=1 and n=2 bands could be dynamically represented by vortex-Rossby waves. On the observational side, Reasor et al. (2000) found close correlation between n=2 bands of vorticity and reflectivity in pseudo-Dual-Doppler analyses of the inner core of Hurricane Olivia (1998). Very strong support for the predictions of vortex-Rossby wave theory in regards to the azimuthal and radial propagation of the bands was recently published by Corbosiero et al. (2006). They performed Fourier decompositions in the azimuthal direction of the radar features, relative to the centre of the storm. The n=2 components of the reflectivity were shown to propagate azimuthally (retrograding relative to the mean flow) and radially (outward) at speeds consistent with the phase and group speeds derived by Montgomery and Kallenbach (1997) and Moeller and Montgomery (2000). For higher azimuthal wavenumbers (n=3 and higher), PV features, convection, and precipitation become less satisfyingly correlated, and vortex-Rossby wave theory is less successful in predicting their behaviour. This may be for several reasons. First, on smaller scales, the dynamics of the waves may be more heavily influenced by their coupling to convection; that is to say, the PV may be a “slave” to the precipitating band, rather than the other way around. Secondly, the inherent noisiness of a highly nonlinear fluid dynamical system with embedded nonlinear physical processes may make the correlations difficult to establish on smaller space and time scales. Finally, smaller-scale bands may have a different dynamical mechanism. This will now be discussed. 1.2.4.2 Smaller scale and “fine-scale” bands In a widely cited paper, Gall et al. (1998) presented detailed analyses of land-based Doppler radar observations of Hugo (1989), Andrew (1992), and Erin (1995). Along with the inner core bands, they also found considerable evidence for numerous, even prolific bands on still smaller scales of 2-10 km, mostly within close range of the eyewall (Fig 1.2.7a). Unlike the larger inner-core bands, they found very little propagation of these smaller bands relative to the mean flow (even less than would be predicted by vortex-Rossby wave theory). They referred to these features as “fine-scale” spiral bands. Very recently, Kusunoki and Mashiko (2006) presented new observations of fine-scale bands, very similar to those of Gall et al. (1998), from radar observations of Typhoon Songda (2004) during its landfall at Okinawa. In a high resolution (2 km) simulation of Hurricane Andrew, Yau et al. (2004) found fine-scale bands with horizontal scales and vertical structures very similar to those shown by Gall et al. (1998) (Fig 1.2.7b). The bands represented a high degree of, but not complete, correlation between PV, vertical velocity, precipitation, and low-level wind maxima. A budget analysis of the wind field showed that the wind streaks were caused by enhanced radial advection of the azimuthal wind field, in turn caused by enhanced vertical motion associated with latent heat release in the bands. However, an analysis of the propagation speed and evolution of these bands as compared to the predictions of vortex-Rossby wave theory was not presented in this study. Nolan (2005) proposed that at least some of these features are not caused by radiating PV waves, but instead are convective bands triggered by large rolls in the hurricane boundary layer. His linearised,

99

global stability analysis of an unstratified swirling boundary layer, modelled after tropical cyclones, did find unstable modes in the form of streamwise rolls with similar scales to the fine-scale bands, that could spiral either inward, outward, or not at all (Fig 1.2.7d,e). The dynamics of these modes are essentially the same as the more common rolls in the Ekman-like planetary boundary layer, but with enhanced instability due to the much larger cross-stream component of the flow (the radial inflow and its reversal above the boundary layer). Foster (2005) treated the problem locally, allowing for stratification and more realistic wind profiles, and also found streamwise rolls (Fig 1.2.7f,g) on a wide range of scales, from those of the fine-scale bands to much smaller rolls embedded in the lowest levels of the boundary layer, as discussed in section 1.2.2.2. In a further attempt to identify and understand the fine-scale bands, Romine and Wilhelmson (2006) analysed a very high resolution (1.1 km inner nest) simulation of Hurricane Opal (1995). The model reproduced the prolific small-scale bands in impressive detail (Fig 1.2.7c). Based on examination of the bands and larger flow field, Romine and Wilhelmson argue that the bands are a manifestation of Kelvin-Helmholtz instability associated with the strong low-level shear of the radial inflow, but modified by gravity wave dynamics and vertical variations of the stability parameter so as to acquire significant outward propagation. However, a more comprehensive analysis of the flow stability, as in Nolan (2005) or Foster (2005), or of the wave dynamics, as in Chen et al. (2003), will be required to validate this hypothesis. 1.2.4.3 Band features and dynamics associated with gravity waves While some of the earliest theories of spiral bands attributed their existence to gravity waves embedded in a swirling flow (Diercks and Anthes 1976; Kurihara 1976; Willoughby 1978), such theories have not been supported by observations and simulations. The main disagreement is that spiral rainband features move quite slowly (or not at all) relative to the symmetric flow, while gravity waves propagate outward quite rapidly, with almost no interaction with the symmetric vortex (Nolan and Montgomery 2002). Nonetheless, radiating gravity waves are endemic to tropical cyclone simulations, as shown by Chow et al. (2002). They are radiated from the eyewall and inner-core region by either isolated convective events or by convective asymmetries propagating on the eyewall, which themselves are often associated with trapped vortex-Rossby waves or even unstable modes. Chow et al. attribute the moving outer spiral rainbands to these waves, while Wang (2002a, b) showed that the outer rainbands (outside of about three times of the RMW) couldn’t be explained by vortex-Rossby waves since the radial PV gradients become too weak to support them. Chow and Chan (2003) derived an Eliassen-Palm-Flux theory for radiating gravity waves in a shallow-water vortex, from which they estimated that gravity waves transport a large amount of angular momentum away from the core, as much as 10% per hour. However, this large loss of angular momentum cannot be reconciled with numerous other studies. Studies of the effects of asymmetric convection on tropical cyclones have found that virtually all of the dynamical effects are localized to the vicinity of the convective event, while any change to the wind field caused solely by the radiating gravity waves is insignificant in comparison (Nolan and Montgomery 2002; Nolan and Grasso 2003; Nolan et al. 2006).

100

Figure 1.2.7: Observed, simulated, and predicted structures of fine-scale bands in various recent publications: a) radius-height radar cross-section of bands in Hurricane Andrew (1992) from Gall et al. (1998); b) radius-height cross-section of simulated reflectivity in a high-resolution simulation of Andrew from Yau et al. (2004); c) vertical motion in a vertical cross-section of a simulation of Hurricane Opal (1995) from Romine and Wilhelmson (2006); d) vertical motion as seen in highly idealized simulations on a hurricane-like boundary layer from Nolan (2005); e) analyzed structure of the global, most unstable mode of the same simulation as d; f) analyzed along-roll velocity of a local unstable mode in a more realistic boundary layer (with stratification) from Foster (2005); g) vertical velocity of the same mode as f.

101

1.2.4.4 RAINEX In the Atlantic hurricane season of 2005, a very ambitious field project was undertaken by lead investigators from the University of Miami and the University of Washington, along with collaborators from NOAA/HRD, NCAR, and the U.S. Navy: the Rainband and Intensity Change Experiment (RAINEX). The goal of the project was the simultaneous observation of the eyewall, inner core, and inner core bands of mature tropical cyclones, with the purpose of trying to determine the role that rainbands play in tropical cyclone intensity changes. An ambitious aspect of the project was the use of three aircraft with on-board Doppler radars at the same time: the two NOAA P3 aircraft and the Navy P3 aircraft. Furthermore, real-time forecasts of the storms using an ensemble of mesoscale model runs (MM5 and WRF, run from various global model forecasts such as the GFS, NOGAPS, and CMC) were used to make daily decisions for flight operations. Thanks in part to the amazingly high activity of the 2005 hurricane season, the observational phase of the project was an outstanding success. In particular, rainbands in the cores of mature hurricanes Katrina, Ophelia, and Rita were sampled extensively and repeatedly by multiple aircraft. In some cases, bands were observed from both sides at the same time, while in others, simultaneous eyewall evolution and rainband evolution were documented. Some noteworthy examples of observed features are the variation of the convective structure along the length of the bands, with convective cells preferred upwind/outward and stratiform precipitation preferred downwind/inward along the band, and detailed wind fields, vorticity, and secondary circulations in the vicinity of a developing secondary eyewall. Analysis of the enormous reflectivity, Doppler wind, and dropsonde data sets from the RAINEX project have just begun. We expect that numerous publications will result between now and the next IWTC report. 1.2.4.5 Rainband activity and organization in regards to the TC environment The organization of rainbands around the storm is known to be highly dependent on the surrounding environment. It as already well know that wind shear has a strong influence, with outer rainbands preferring the downshear right quadrant, often organizing into a single, long lasting “principal band” that leads the storm (Willoughby et al. 1984). Inner-core bands are also suppressed in the upshear quadrants. The organization of convective activity in regards to storm motion, vertical shear direction, and storm strength was documented by Lonfat et al. (2004) with composites of TRMM satellite rain rate observations from numerous storms around the world. Two very recent studies have independently identified the size and strength of the “moisture envelope” around the storm as playing a significant role in organizing the bands. Using RAINEX and global satellite data to compare Hurricanes Katrina (2005) and Rita (2005), Ortt and Chen (2006) found that a large envelope of above average moisture around Katrina caused rainband formation over a very large area around the storm. A smaller moist envelope around Rita, however, suppressed the outer rainband activity, allowing for a more focused region of more intense inner core band activity in close vicinity of the storm. This in turn led to multiple eyewall replacement cycles for Rita, while only one such cycle occurred for Katrina as it traversed the gulf. Similar changes in rainband organization were found by Kimball (2006) in mesoscale model simulations of the landfall of Hurricane Danny (1997). The specific humidity of the environment around the storm was modified with large and small-sized areas of enhanced and suppressed moisture. Larger and stronger moist envelopes led to enhanced rainband activity around the storm. The effects of these changes on storm intensity are discussed in the next section.

102

1.2.4.6 The effect of rainbands on hurricane intensity Does the existence of spiral rainbands make hurricanes stronger, weaker, or have little effect? This is the over-arching question in hurricane research that focuses on spiral bands. In past years spiral bands were seen as having a negative impact on intensity, as they steal some of the low level moist inflow into the eye of the storm, while simultaneously replacing that inflow with cooler, drier air from convective downdrafts (Barnes et al., 1983; Powell, 1990a, 1990b; Cione et al. 2000; Wang, 2002c; Wroe and Barnes, 2003). However, the introduction of cooler and drier air allows for more heat and moisture to be extracted from the ocean by the storm. Nonetheless, this extra heat and moisture simply replaces what is released in the rainbands, and latent heat release well away from the storm centre is believed to have very little effect on intensity (Hack and Schubert, 1986; Nolan et al., 2006). From another thermodynamic point of view, however, rainbands may help maintain storm intensity in the face of shear and/or dry air associated with approaching troughs or jets. It is widely believed that “large” storms can resist the effects of shear and dry air for a longer time before the storm begins to weaken. This is in agreement with the simulations of Kimball (2006) described above. Kimball also observed that a “large-moist-envelope” storm with enhanced rainband activity intensified more slowly. Thus, the thermodynamic effect of rainbands on a tropical cyclone likely depends on the stage of development of the storm. There is ongoing research about the effects of rainbands on cyclone intensity from a fluid dynamical point of view. Spiral bands of convection seem to both be caused by, and to generate, spiral bands of PV. These spiral features are examples of vorticity perturbations that are tilted “downshear” in a stable, sheared flow. Such perturbations are universally associated with an “upgradient” (in this case, inward) transport of angular momentum, leading to a small increase in the intensity of the storm (Carr and Williams 1989; Montgomery and Kallenbach 1997; Nolan and Farrell 1999b; Moeller and Montgomery 1999, 2000; May and Holland 1999; Chen et al. 2003). However, recent results by Nolan and Grasso (2003) and Nolan et al. (2006) suggest this conclusion may be incomplete. By simulating the generation of vortex-Rossby waves from isolated convective events embedded in a vortex, they found that while the waves themselves cause an intensifying effect on the vortex, the adjustment process whereby the bands were created had an even larger negative effect. In any case, the net effect of the “asymmetric” dynamics was trivial compared to the effect of the azimuthally averaged heating. Thus, while spiral rainbands may be associated with inward transport of angular momentum and vorticity, their net effect on the storm is more likely dominated by the azimuthally averaged secondary circulation generated by their associated convection and latent heat release. As before, then, the overall impact of spiral rainbands on tropical cyclone intensity cannot be stated in generalities. Recent advances, however, are leading to a broader understanding from which the effects of rainbands can be inferred on a case-by-case basis. 1.2.5. Implications for Forecasting Our ability to accurately forecast hurricane motion (track) has improved dramatically in the past 20 years, largely because of the improving ability to capture evolving synoptic-scale fields with present numerical guidance. The track is controlled almost entirely by the environmental steering flow in which the storm vortex is embedded, and which is increasingly well depicted in global- and regional-scale NWP. In contrast, our ability to forecast hurricane intensity change has shown quite limited progress in the past 20 years and again it is widely believed that this is due to our present inability to model small-scale internal processes in hurricanes. Here, we discuss the implications of the recent gains in our knowledge of inner-core structure and dynamics, as reviewed in previous sections, on the

103

forecasting of intensity and track. 1.2.5.1 Inner core influences on forecasting intensity and intensity change Intensity, when measure in terms of near-surface wind speed, is determined partly by the properties of the boundary layer. The boundary layer has been shown by theory, models and observations to be very inhomogeneous, particularly in the vicinity of the eyewall. The most direct impact of this for intensity estimation and forecasting is that the surface wind factor is higher beneath the eyewall than at larger radius (0.9 instead of 0.8), and is frequently higher on the weak (left in the Northern Hemisphere) side of the storm. Measurements of near-surface turbulence have tended to confirm existing gust factors. Strong evidence for boundary layer rolls explains the observed streakiness of damage in some storms, and may indicate higher risk than has hitherto been supposed. The eyewall replacement cycle has been known for decades, but satellite microwave observations continue to improve our ability to monitor it, and to make short-term predictions by extrapolation. Such ability is particularly high at the moment, due to the large number of satellites in operations, and unlikely to be sustained. Mesoscale vortices on the inner edge of the eyewall have been shown to be associated with locally enhanced winds. They thus present an additional hazard, as well as playing a key role in cyclone dynamics. While observations have generally been in very vertical storms, there is some evidence that they also occur in tilted systems. Their hypothesised role in the formation of “annular hurricanes” is significant to forecasting, since such storms have been shown to decay at significantly less than the climatological rate. Moreover, the fact that they can cause significant rearrangements of the inner core structure, including weaker winds with lower central pressure, casts doubt on the long-established practice of regarding maximum wind and central pressure as quasi-equivalent measures of storm intensity. Rainbands continue to have an uncertain impact on TC intensity: they consume some of the energy-rich low-level moist inflow, but may protect the inner core from environmental shear, and also produce intensification through the transport of angular momentum into the storm centre. Numerical weather prediction (NWP) of these changes remains challenging. Many of the relevant features are short-lived and evolve rapidly. Their initiation may depend on small-scale, poorly-observed features in the storm. The dynamics may be so different to much of the rest of the atmosphere, and the physics of the observations so complicated (e.g. ice scattering in passive microwave satellite imagery) that data assimilation systems are stretched to interpret it. The time-scale may also be much less than the traditional 6-hourly NWP cycle. Nevertheless, a few encouraging preliminary results have been achieved. Intensity change is also strongly influenced by external forcing, either from the atmospheric environment surrounding the storm, from the ocean, or from nearby land and orography. These are covered in detail in subtopics 1.1 and 1.3. Subtopic 1.5 focuses directly on the operational forecasting of structure change. 1.2.5.2 Inner core influences on forecasting track In contrast to intensity, the influence of the inner core on forecasting track is modest. Small-scale asymmetries can give rise to a trochoidal oscillation (e.g. Muramatsu, 1986; Nolan et al., 2001), but the small amplitude relative to the scale of the storm, and short time-scale of the oscillation relative to operational cycles, means that this will have only slight operational significance.

104

1.2.6. Conclusions Better observations and theoretical advances have resulted in significant improvements in understanding. A short summary of the highlights follows:

• There have been significant advances in knowledge and understanding of surface winds on several scales: the mean, coherent boundary-layer structures, and turbulence.

• The winds in the upper boundary-layer are supergradient in some storms.

• Multiple satellites and radars are providing an unprecedented view of the eyewall replacement cycle, shedding new light on its occurrence and importance in storm dynamics. These observations are informing operational forecasts of intensity change, on occasion.

• Eyewall mesovortices provide an added hazard in some storms, and play an important role in storm dynamics and intensity change.

• Vortex Rossby waves explain observations of inner-core spiral bands, and play an important role in the storm dynamics, including in genesis, resistance to shear, and intensification. Fine-scale bands may be from other causes, including boundary layer instabilities.

1.2.7. Recommendations for Research and Operations Boundary layer winds. How much of the inter- and intra-storm variability in surface wind factor is predictable? How does this compare with other sources of uncertainty in intensity estimation? How important is stability? What are suitable gust factors over the sea/land/beneath the eyewall/at larger radius? Is it necessary/possible to predict the occurrence/intensity of rolls? Wind-field expansion. This is clearly operationally important (e.g. H. Katrina’s surge) but little (nothing?) is known of the dynamics. How well can we predict this? Eyewall mesovortices. How prevalent are these? In vertical/tilted storms? How often are they sufficiently intense to represent a significant additional hazard? Can their formation be predicted (for application to intensity change, annular hurricanes)? Spiral bands. In the inner core, vortex Rossby waves are an important, but not the sole, influence. But at larger radius? What is the role of gravity waves? And the implication for numerical modelling? Do the wind perturbations associated with fine-scale inner bands matter for risk estimation? Do spiral bands intensify or weaken the storm, or does it depend on the situation? If so, how? Numerical Weather Prediction. Which of these small-scale features can be predicted? How far ahead? What are the issues for assimilation of various data types in tropical cyclones? Do assimilation systems contain the right balances for tropical cyclones? Empirical Prediction. How can knowledge of these processes improve forecasts? Especially in the last few hours before landfall when they may be visible on radar as well as satellite? Can Doppler radar algorithms (e.g. EVTD) be used in real-time? How can these be shared around the world? How will emergency services respond to such “last-minute” information? What are the limits on intensification/decay rates? What are the precursors of intensification/decay on various types of satellite imagery? How close must land/islands be before they start to affect structure?

105

Acronyms AMSR-E Advanced Scanning Microwave Radiometer for EOS BL Boundary-layer EMV Eyewall mesovortex ERC Eyewall replacement cycle EVTD Extended Velocity Track Display IR Infrared MHS-1 Microwave Humidity Sounder NHC National Hurricane Center NRC National Research Council PI Potential intensity PV Potential vorticity RMW Radius of maximum (tangential?) wind speed SFMR Stepped-frequency microwave radiometer SSM/I Special Sensor Microwave Imager SSMIS Special Sensor Microwave Imager/Sounder SST Sea surface temperature SWF Surface wind factor TC Tropical cyclone TCBL Tropical cyclone boundary layer TMI TRMM Microwave Imager TRMM Tropical Rainfall Measuring Mission Bibliography Note that most, but not all these references are cited in the text - the bibliography is intended to be as exhaustive as we could make it. Aberson, S.D. and J.B. Halverson. 2006: Kelvin–Helmholtz billows in the eyewall of Hurricane Erin. Mon. Wea. Rev., 134, 1036–1038. Aberson, S.D., J.P. Dunion and F.D. Marks Jr., 2006: A photograph of a wavenumber-2 asymmetry in the eye of Hurricane Erin. J. Atmos. Sci., 63, 387–391. Aberson, S.D. and D.P. Stern, 2006: Extreme horizontal winds measured by dropwindsondes in hurricanes. Extended Abstracts, 27th Conference on Hurricanes and Tropical Meteorology, Amer. Meteor. Soc., Monterey, Ca, 24 - 28 April. CD-ROM. Barnes, G.M., E.J. Zip[ser, D.P. Jorgensen and F.D. Marks Jr., 1983: Mesoscale and convective structure of a hurricane rainband. J. Atmos. Sci., 40, 2125-2137. Bender, M.A., 1997: The effect of relative flow on the asymmetric structure in the interior of hurricanes. J. Atmos. Sci., 54, 703 - 724.

Bister, M., 2001: Effect of Peripheral Convection on Tropical Cyclone Formation. J. Atmos. Sci. 58, 3463-3476. Black, M.L.; Gamache, J.F.; Marks, F.D. Jr; Samsury, C.E.; Willoughby, H.E., 2002: Eastern Pacific Hurricanes Jimena of 1991 and Olivia of 1994: The Effect of Vertical Shear on Structure and Intensity. Mon. Weather Rev. 130, 2291-2312

106

Black, P. G., and F. D. Marks, 1991: The structure of an eyewall meso-vortex in Hurricane Hugo (1989). Preprints, 19th Conf. on Hurricanes and Tropical Meteorology, Miami, FL, Amer. Meteor. Soc., 579–582.

Braun, S.A., 2002: A Cloud-Resolving Simulation of Hurricane Bob (1991): Storm Structure and Eyewall Buoyancy. Mon. Wea. Rev., 130, 1573-1592. Braun, S.A., M.T. Montgomery and Z. Pu. 2006: High-Resolution Simulation of Hurricane Bonnie (1998). Part I: The organization of eyewall vertical motion. J. Atmos. Sci., 63, 19–42. Brown, R. A., and L. Zeng, 2001: Comparison of Planetary Boundary Layer Model Winds with Dropsonde Observations in Tropical Cyclones. J. AMeteor., 40, 1718-1723. Camp, J. P., and M. T. Montgomery, 2001: Hurricane Maximum Intensity: Past and Present. Mon. Wea. Rev., 129, 1704-1717. Cangialosi, J.P., S.S. Chen, W. Zhao, W. Wang, and J. Michalakas, 2006: Real-time high-resolution MM5 and WRF forecasts during RAINEX. Extended Abstracts, 27th Conference on Hurricanes and Tropical Meteorology, Amer. Meteor. Soc., Monterey, Ca, 24 - 28 April. CD-ROM. Carr, L.E. III and R.T. Williams. 1989: Barotropic Vortex Stability to Perturbations from Axisymmetry. J. Atmos. Sci., 46, 3177–3191. Cerveny, Randall S.; Balling, Robert C. 2005: Variations in the Diurnal Character of Tropical Cyclone Wind Speeds. Geophys. Res. Lett.: 32, L06706. Chen, Y., and M. K. Yau, 2001: Spiral Bands in a Simulated Hurricane. Part I: Vortex Rossby Wave Verification. J. Atmos. Sci., 58, 2128-2145. Chen, Y., G. Brunet, and M. K. Yau. 2003: Spiral Bands in a Simulated Hurricane. Part II: Wave Activity Diagnostics. J. Atmos. Sci.. 60, 1239-1256. Chow, K. C., Chan, K. L., Lau, A.K.H.. 2002: Generation of Moving Spiral Bands in Tropical Cyclones. J. Atmos. Sci.. 59, 2930–2950. Chow, K.C. and K.L. Chan, 2003: Angular momentum transports by moving spiral waves. J. Atmos. Sci., 60, 2004-2009. Cione, J.J., P.G. Black and S.H. Houston, 2000: Surface wind observations in the hurricane environment. Mon. Wea. Rev., 128, 1550-1561. Corbosiero, K., and J. Molinari, 2002: The effects of vertical wind shear on the distribution of convection in tropical cyclones. Mon. Wea. Rev., 130, 2110-2123. Corbosiero, K., and J. Molinari, 2003: The relationship between storm motion, vertical wind shear, and convective asymmetries in tropical cyclones. J. Atmos. Sci., 60, 366-376. Corbosiero, Kristen L., John Molinari and Michael L. Black. 2005: The Structure and Evolution of Hurricane Elena (1985). Part I: Symmetric Intensification. Mon. Wea. Rev.: 133, 2905–2921 Corbosiero, K. L., J. Molinari, and A. R. Aiyyer, 2006: The structure and evolution of Hurricane Elena (1985). Part II: Convective asymmetries and evidence for vortex-Rossby waves. Mon. Wea. Rev., in press. Cram, T. A., J. A. Knaff, M. DeMaria, and J. P. Kossin, 2006, Objective identification of annular

107

hurricanes using GOES and reanalysis data, AMS 27th Conf. on Hurricanes and Tropical Meteorology, CD-ROM. Croxford, M., and G. M. Barnes, 2002: Inner core strength of Atlantic tropical cyclones. Mon. Wea. Rev., 130, 127-139. Dengler K., and D. Keyser (2000) Intensification of tropical cyclone-like vortices in uniform zonal background flows. Quart. J. Roy. Meteor. Soc., 126, 549–568. Diercks, J.W. and R.A. Anthes. 1976: A Study of Spiral Bands in a Linear Model of a Cyclonic Vortex. J. Atmos. Sci., 33, 1714–1729. Dunion, J.P., C.W. Landsea, S.H. Houston, And M.D. Powell, 2003: A Reanalysis of the Surface Winds for Hurricane Donna of 1960, Mon. Wea. Rev., 131, 1992-2011.

Eastin, M.D., 2002. Observational analysis of buoyancy in intense hurricane eyewalls. Extended abstracts, 24th Conference on Hurricanes and Tropical Meteorology, American Meteorological Society, Boston, pp. 640–641.

Eastin, M.D., 2003. Buoyancy of convective vertical motions in the inner core of intense hurricanes. Ph.D. dissertation, Department of Atmospheric Science, Colorado State University, Fort Collins, Colorado 80521, USA, p. 152. Eastin, M.D., W.M. Gray and P.G. Black. 2005a: Buoyancy of Convective Vertical Motions in the Inner Core of Intense Hurricanes. Part I: General Statistics. Mon. Wea. Rev.: 133, 188–208 Eastin, M.D., W.M. Gray and P.G. Black. 2005b: Buoyancy of Convective Vertical Motions in the Inner Core of Intense Hurricanes. Part II: Case Studies. Mon. Wea. Rev.: 133, 209–227 Eliassen, A. and M. Lystad, 1977: The Ekman layer of a circular vortex: A numerical and theoretical study. Geophysica Norvegica, 31, 1-16.

Emanual, K.A., 1986: An air-sea interaction theory for tropical cyclones. Part I: Steady-state maintenance. J. Atmos. Sci., 43, 585-604.

Emanuel, K.A., 1991: The theory of hurricanes. Ann. Rev. Fluid Mech., 23, 179-196.

Emanuel, K.A., 1995: Sensitivity of tropical cyclones to surface exchange coefficients and a revised steady-state model incorporating eye dynamics. J. Atmos. Sci., 52, 3969-3976.

Emanuel, K.A., 2000: A statistical analysis of tropical cyclone intensity. Mon. Wea. Rev., 128, 1139 - 1152.

Emanuel, Kerry, C. DesAutels, C. Holloway and R. Korty. 2004: Environmental Control of Tropical Cyclone Intensity. J. Atmos. Sci., 61, 843-858.

Etling, D., and R.A. Brown, 1993: Roll vortices in the planetary boundary layer: A review. Boundary Layer Meteorology, 65, 215-248.

Evans, J.L. and R. Hart, 2003: Objective indicators of the extratropical transition lifecycle of Atlantic tropical cyclones. Mon. Wea. Rev., 131, 909-925.

Farfán, L. M., and J. A. Zehnder, 2001: An Analysis of the Landfall of Hurricane Nora (1997). Mon. Wea. Rev., 129, 2073-2088.

108

Foster, R.C., 2005: Why Rolls are Prevalent in the Hurricane Boundary Layer. J. Atmos. Sci.: 62, 2647–2661 Frank, W.M., and E.A. Ritchie, 1999: Effects of environmental flow upon tropical cyclone structure. Mon. Wea. Rev., 127, 2044-2061. Frank, W. M., and E. A. Ritchie, 2001: Effects of Vertical Wind Shear on the Intensity and Structure of Numerically Simulated Hurricanes. Mon. Wea. Rev., 129, 2249-2269. Franklin, J.L., M.L. Black and K. Valde, 2003: GPS dropwindsonde wind profiles in hurricanes and their operational implications. Wea. and Forecasting, 18, 32-44.

Free, Melissa, M. Bister and K. Emanuel. 2004: Potential Intensity of Tropical Cyclones: Comparison of Results from Radiosonde and Reanalysis Data. Journal of Climate: 17, 1722-1727. Gall, R., J. Tuttle, and P. Hildebrand, 1998: Small-scale spiral bands observed in hurricanes Andrew, Hugo, and Erin. Mon. Wea. Rev., 126, 1749-1766.

Hack, J.J. and W.H. Schubert. 1986: Nonlinear Response of Atmospheric Vortices to Heating by Organized Cumulus Convection. J. Atmo. Sci., 43, 1559–1573.

Halverson, J. B., J. Simpson, G. Heymsfield, H. Pierce, T. Hock and L. Ritchie. 2006: Warm core structure of Hurricane Erin diagnosed from high altitude dropsondes during CAMEX-4. J. Atmos. Sci., 63, 309–324. Hanley, D., J. Molinari and D. Keyser. 2001: A Composite Study of the Interactions between Tropical Cyclones and Upper-Tropospheric Troughs. Mon. Wea. Rev., 129, 2570–2584.

Hart, R.,2003: A cyclone phase space derived from thermal wind and thermal asymmetry. Mon. Wea. Rev., 131, 585-616. Hausman, S.A., K.V. Ooyama and W.H. Schubert. 2006: Potential vorticity structure of simulated hurricanes. J. Atmos. Sci., 63, 87–108. Hawkins, J. D., T. F. Lee, K. Richardson, C. Sampson, F. J. Turk, and J. E. Kent, 2001, Satellite multi-sensor tropical cyclone structure monitoring, Bull. Amer. Meteor. Soc., 82, 567-578. Hawkins, J. D., and M. Helveston, 2004, Tropical cyclone multi-eyewall characteristics, Preprints AMS 26th Hurricane and Tropical Meteorology Conference, 276-277. Hawkins, J. D., M. Helveston, T. F. Lee, F. J. Turk, K. Richardson, C. Sampson, J. Kent and R. Wade, 2006, Tropical cyclone multiple eyewall configurations, 27th AMS Conference on Hurricanes and Tropical Meteorology, CD-ROM. Hendricks, E.A.; Montgomery, M.T.; Davis, C.A., 2004: The Role of 'Vortical' Hot Towers in the Formation of Tropical Cyclone Diana (1984). J. Atmos. Sci. 61, 1209-1232.

Henning, R.G. , 2006: The intensity of wind gust underneath areas of deep eyewall convection in Hurricanes Katrina and Dennis at landfall. Extended Abstracts, 27th Conference on Hurricanes and Tropical Meteorology, Amer. Meteor. Soc., Monterey, Ca, 24 - 28 April. CD-ROM.

Heymsfield, G. M., J. B. Halverson, J. Simpson, L. Tian, and T. P. Bui, 2001: ER-2 Doppler Radar Investigations of the Eyewall of Hurricane Bonnie during the Convection and Moisture Experiment-3. J. AMeteor., 40, 1310-1330.

109

Heymsfield, G.M., Halverson, J, Ritchie, E.A., Simpson, J, Molinari, J., Tian, L., 2006: Structure of highly sheared Tropical Storm Chantal during CAMEX-4. J. Atmos. Sci. 63, 268-287. Hock, T.F. and J.L. Franklin, 1999: The NCAR GPS dropwindsonde. Bull. Amer. Meteor. Soc., 80, 407-420.

Holland, G.J., 1980: An analytic model of the wind and pressure profiles in hurricanes. Mon. Wea. Rev., 108, 1212-1218.

Holland, G.J. and R.T. Merrill, 1984: On the dynamics of tropical cyclone structural changes. Quart. J. Roy. Meteorol. Soc., 110, 723-745.

Holland, G.J, and M.Lander, 1993: The meandering nature of tropical cyclone tracks. J. Atmos. Sci., 50, 1254-1266. Hollinger, J., 1989, DMSP Special Sensor Microwave/Imager calibration /validation. Final Rep. Vol. 1. Naval Research Laboratory, 153 pp. [Available from the Naval Research Laboratory, Washington, DC, 20375]. Horstmann, J.; Thompson, D.R.; Monaldo, F.; Iris, S.; Graber, H.C. 2005: Can Synthetic Aperture Radars be Used to Estimate Hurricane Force Winds? Geophys. Res. Lett.: 32, L22801 Hoskins, B.J., M.E. McIntye, and A.W. Robertson, 1985: On the use and significance of isentropic potential vorticity maps. Q. J. Roy. Meteor. Soc., 121, 821-851. Jones, S.C., 1995: The evolution of vortices in vertical shear. Part I: Initially barotropic vortices. Quart. J. Roy. Meteorol. Soc., 121, 821 - 851.

Jones S.C., 2000: The evolution of vortices in vertical shear. II: Large-scale asymmetries. Quart. J. Roy. Meteor. Soc., 3137-3160. Jones S.C., 2000: The evolution of vortices in vertical shear. III: Baroclinic vortices. Quart. J. Roy. Meteor. Soc., 3161-3186. Jones, Sarah C. 2004: On the Ability of Dry Tropical-Cyclone-like Vortices to Withstand Vertical Shear. J. Atmos. Sci.: 61, 114-119. Kaplan, John, DeMaria, Mark. 2003: Large-Scale Characteristics of Rapidly Intensifying Tropical Cyclones in the North Atlantic Basin. Weather and Forecasting. 18, 1093–1108. Katsaros, K. B., P. W. Vachon, W. T. Liu, and P. G. Black, 2002:Microwave remote sensing of tropical cyclones from space. J. Oceanogr., 58, 137–151.

Kelley, Owen A.; Stout, John; Halverson, Jeffrey B. 2005: Hurricane Intensification Detected by Continuously Monitoring Precipitation in the Eyewall. Geophys. Res. Lett.: 32, L20819. Kepert, J. D., 2001: The Dynamics of boundary layer jets within the tropical cyclone core. Part I : Linear Theory J. Atmos. Sci., 58, 2469-2484. Kepert, J. D. and Y. Wang, 2001: The Dynamics of boundary layer jets within the tropical cyclone core. Part II : Nonlinear enhancement J. Atmos. Sci., 58, 2485-2501. Kepert, J.D., 2002a: The impact of landfall on tropical cyclone boundary layer winds. Extended

110

abstracts, 25th Conference on Hurricanes and Tropical Meteorology, Amer. Meteor. Soc., San Diego, California, 29 April - 3 May, 2002, 335-336. Kepert, J.D., 2002b: Modelling the tropical cyclone boundary layer wind-field at landfall. Extended abstracts, 14th BMRC Modelling Workshop: Modelling and Predicting Extreme Events, Melbourne, Australia, 11-13 November, 81-84. Kepert, J.D., 2002c: The Wind-Field Structure of the Tropical Cyclone Boundary-Layer. PhD thesis, Monash University, Melbourne, Australia. 350 pp. Kepert, J.D. 2004: Models and observations of tropical cyclone boundary-layer winds. BMRC Research Report No 104: The Past, Present and Future of Numerical Modelling: extended abstracts of presentations at the sixteenth annual BMRC Modelling Workshop 6-9 December 2004. Melbourne, Australia. pp 77-82.

Kepert, Jeffrey D., 2005: Objective Analysis of Tropical Cyclone Location and Motion from High-Density Observations. Mon. Wea. Rev.: 133, 2406–2421. Kepert, J.D., 2006a: Observed boundary-layer wind structure and balance in the hurricane core. Part I: Hurricane Georges. In press, J. Atmos. Sci. Kepert, J.D., 2006b: Observed boundary-layer wind structure and balance in the hurricane core. Part II: Hurricane Mitch. In press J. Atmos. Sci. Kim, S.Y.; Chun, H.Y.; Baik, J.J. 2005: A Numerical Study of Gravity Waves Induced by Convection Associated with Typhoon Rusa. Geophys. Res. Lett.: 32, L24816. Kimball, S.K., 2006: A modelling study of hurricane landfall in a dry environment. Mon. Wea. Rev., 134, 1901-1918. Kishtawal, C.M.; Patadia, Falguni; Singh, Randhir, Basu, Sujit; Narayanan, M.S.; Joshi, P.C. 2005: Automatic Estimation of tropical Cyclone Intensity Using Multi-Channel TMI Data: A genetic Algorithm Approach. Geophys. Res. Lett.: 32, L11804. Knaff, J. A., J. P. Kossin, and M. DeMaria, 2003, Annular hurricanes, Wea. & Forecasting, 18, 204-223. Knaff, John A., S. A. Seseske, M. DeMaria and J. L. Demuth. 2004: On the Influences of Vertical Wind Shear on Symmetric Tropical Cyclone Structure Derived from AMSU. Mon. Wea. Rev.: 132, 2503-2510. Knupp, K.R., J. Walters, and E.W. McCaul Jr, 2000: Doppler profiler observations of Hurricane Georges at landfall. Geophysical Research Letters, 27, 3361-3364.

Knupp K.R., J. Walters, M. Biggerstaff, 2005: Doppler Profiler and Radar Observations of Boundary Layer Variability during the Landfall of Tropical Storm Gabrielle. J. Atmos. Sci., 63, 234-251.

Kodama, Yasu-Masa and Takuya Yamada. 2005: Detectability and Configuration of Tropical Cyclone Eyes over the Western North Pacific in TRMM PR and IR Observations. Mon. Wea. Rev.: 133, 2213–2226 Kodama, Y., T. Yamada, 2005, Detectability and configuration of tropical cyclone eyes over the western north Pacific in TRMM PR and IR observations, Mon. Wea. Rev., 133, 2213-2226. Kossin, J. P., Schubert W. T., Montgomery M. T., 2000: Unstable interactions between a hurricane's

111

primary eyewall and a secondary ring of enhanced vorticity. J. Atmos. Sci. 57, 3893-3917. Kossin, J. P., and M. D. Eastin, 2001: Two Distinct Regimes in the Kinematic and Thermodynamic Structure of the Hurricane Eye and Eyewall. J. Atmos. Sci., 58, 1079-1090. Kossin, J. P., and W. H. Schubert, 2001: Mesovortices, Polygonal Flow Patterns, and Rapid Pressure Falls in Hurricane-Like Vortices. J. Atmos. Sci., 58, 2196-2209. Kossin, J. P., B. D. McNoldy, and W. H. Schubert, 2002: Vortical swirls in hurricane eye clouds. Mon. Wea. Rev., 130, 3144-3149 Kossin, J. P., and W. H. Schubert, 2003: Diffusion versus advective rearrangement of a circular vortex sheet. J. Atmos. Sci.. 60, 586-589. Kossin, J., H. Berger, J. Hawkins, and T. Cram, 2006: Development of a secondary eyewall formation index for improvement of tropical cyclone intensity forecasting. Proceedings, 60th Interdepartmental Hurricane Conference (IHC), Mobile, AL. Kossin, J. P. and Wayne H. Schubert 2005: Mesovortices in Hurricane Isabel. Bull. Amer. Meteorol. Soc.: 85, 151-153. Krishnamurti, T. N., S. Pattnaik, L. Stefanova, T. S. V. Vijaya Kumar, B. P. Mackey, A. J. O’Shay and Richard J. Pasch. 2005: The Hurricane Intensity Issue. Mon. Wea. Rev.: 133, 1886–1912 Kuo, H.-C., R. T. Williams and J.-H. Chen. 1999: A Possible Mechanism for the Eye Rotation of Typhoon Herb. J. Atmos. Sci., 56, 1659–1673.

Kuo, H.-C., L.-Y. Lin, C.-P. Chang and R. T. Williams. 2004: The Formation of Concentric Vorticity Structures in Typhoons. J. Atmos. Sci.: 61, 2722-2734. Kuo, H.L., 1971: Axisymmetric flow in the boundary layer of a maintained vortex. J. Atmos. Sci., 28, 20-41.

Kuo, H.L., 1982: Vortex boundary layer under quadratic surface stress. Boundary-Layer Meteorol., 22, 151-169.

Kurihara, Y., 1976: On the development of spiral bands in a tropical cyclone. J. Atmos. Sci., 33, 940-958.

Kurihara, Y.; Bender, M.A.; Ross, R.J., 1993: An initialization scheme of hurricane models by vortex specification. Mon. Wea. Rev., 121, 2030-2045.

Kurihara, Y.; Bender, M.A; Tuleya, R.E.; Ross, R.J., 1995: Improvements in the GFDL hurricane prediction system. Mon. Wea. Rev., 123, 2791-2801.

Kusunoki, K., 2006: Assessment of the Doppler radar for airport weather (draw) system in Japan as a research tool for studying typhoon. Extended Abstracts, 27th Conference on Hurricanes and Tropical Meteorology, Amer. Meteor. Soc., Monterey, Ca, 24 - 28 April. CD-ROM. Kusunoki, K. and W. Mashiko, 2006: Doppler radar investigations of the inner core of typhoon songda (2004): polygonal/ elliptical eyewalls, eye contraction, and small-scale spiral bands. Extended Abstracts, 27th Conference on Hurricanes and Tropical Meteorology, Amer. Meteor. Soc., Monterey, Ca, 24 - 28 April. CD-ROM. Kwok J.H.Y. and J. C. L. Chan. 2005: The Influence of Uniform Flow on Tropical Cyclone Intensity

112

Change. J. Atmos. Sci.: 62, 3193–3212 Kwon Y.C. and W.M. Frank. 2005: Dynamic Instabilities of Simulated Hurricane-like Vortices and Their Impacts on the Core Structure of Hurricanes. Part I: Dry Experiments. J. Atmos. Sci.: 62, 3955–3973. Lackmann, Gary M. and R. M. Yablonsky. 2004: The Importance of the Precipitation Mass Sink in Tropical Cyclones and Other Heavily Precipitating Systems. J. Atmos. Sci.: 61, 1674-1692. Landsea, Christopher W., J. L. Franklin, C. J. McAdie, J. L. Beven II, J. M. Gross, B. R. Jarvinen, R. J. Pasch, E. N. Rappaport, J. P. Dunion and P. P. Dodge. 2004: A Reanalysis of Hurricane Andrew's Intensity. Bull. Amer. Meteorol. Soc.: 85, 1699-1712. Lee, W.-C., B. J.-D. Jou, P. L. Chang, and S.-M. Deng, 1999: Tropical cyclone kinematic structure retrieved from single Doppler radar observations. Part I: Interpretation of Doppler velocity patterns and the GBVTD technique. Mon. Wea. Rev., 127, 2419-2439. Lee, W.-C., and F. D. Marks Jr., 2000: Tropical cyclone kinematic structure retrieved from single Doppler radar observations. Part II: The GBVTD-simplex center finding algorithm. Mon. Wea.Rev., 128, 1925-1936.

Lee, W.-C., B. J.-D. Jou, P. L. Chang, and F. D. Marks Jr., 2000: Tropical cyclone kinematic structure retrieved from single Doppler radar observations. Part III: Evolution and structure of Typhoon Alex (1987). Mon. Wea.Rev., 128, 3982-4001. Lee, T. F., F. J. Turk, J. D. Hawkins, and K. A. Richardson, 2002, Interpretation of TRMM TMI images of tropical cyclones, Earth Interactions E-Journal, 6, 3. Lewis, B.M. and H.F., Hawkins, 1982: Polygonal eye walls and rainbands in hurricanes. Bull. Amer. Meteor. Soc., 63, 1294-1301. Liang, Xudong and Johnny C. L. Chan. 2005: The Effects of the Full Coriolis Force on the Structure and Motion of a Tropical Cyclone. Part I: Effects due to Vertical Motion. J. Atmos. Sci.: 62, 3825–3830. Lin, Y.-L., D.B. Ensley, S. Chiao and C.-Y. Huang, 2002: Orographic Influences on Rainfall and Track Deflection Associated with the Passage of a Tropical Cyclone. Mon. Wea. Rev., 130, 2929–2950. Liou, Y.-C., T.-C. Chen Wang, W.-C. Lee and Y.-J. Chang. 2006: The retrieval of asymmetric tropical cyclone structures using Doppler radar simulations and observations with the Extended GBVTD Technique. Mon. Wea. Rev., 134,1140–1160. Lonfat, M; Marks, FD; Chen, SS, 2004: Precipitation Distribution in Tropical Cyclones Using the Tropical Rainfall Measuring Mission (TRMM) Microwave Imager: A Global Perspective. Mon. Weather Rev., 132, 1645-1660.

Lorsolo, S., and J. L. Schroeder, 2006: Tower And Doppler Radar Observations From The Boundary Layer Of Hurricanes Isabel (2003) And Frances (2004). Extended Abstracts, 27th Conference on Hurricanes and Tropical Meteorology, Amer. Meteor. Soc., Monterey, Ca, 24 - 28 April. CD-ROM. Mallen, K.J.m M.T. Montgomery and B. Wang, 2005: Reexamining the near-core radial structure of the tropical cyclone primary circulation: Implications for vortex resiliency. J. Atmos. Sci., 62, 408-425. Mallett, W.A., 2000: The Dynamics of the Atmospheric Boundary Layer of a Mature Tropical Cyclone: A Numerical Investigation. PhD Thesis, James Cook University, Australia, 266 pp.

113

Marks, F.D., and P. G. Black, 1990: Close encounter with an intense mesoscale vortex within Hurricane Hugo (September 15, 1989. Extended Abstracts, Fourth Conf. on Mesoscale Processes, Boulder, CO, Amer. Meteor. Soc., 114–115.

Marks, F.D. P. Dodge and C. Sandin, 1999: WSR-88D Observations of hurricane atmospheric boundary layer structure at landfall. Extended abstracts, 23rd Conference on Hurricanes and Tropical Meteorology, Amer. Meteorol. Soc., Dallas, TX, Jan. 10-15. 1051-1055.

May, P.T. 1996: The organization of convection in the rainbands of Tropical Cyclone Laurence. Mon. Wea. Rev., 124, 807-815.

May, P.T. and G.J. Holland, 1999: The role of potential vorticity generation in tropical cyclone rainbands. J. Atmos. Sci., 56, 1224-1228.

McNoldy, B. D., 2004, Triple eyewall in hurricane Juliette, Bull. Amer. Meteor. Soc., 85, 1663-1666. Möller, J.D. and M.T. Montgomery. 1999: Vortex Rossby Waves and Hurricane Intensification in a Barotropic Model. J. Atmos. Sci., 56, 1674–1687. Moeller, J. D., and M. T. Montgomery, 2000: Tropical cyclone evolution via potential vorticity anomalies in a three- dimensional balance model. J. Atmos. Sci., 57, 3366-3387. Möller, J. D., and L. J. Shapiro, 2002: Balanced contributions to the intensification on hurricane Opal as diagnosed from a GFDL model forecast. Mon. Wea. Rev., 130, 1866-1881. Möller, J.D. and L. J. Shapiro. 2005: Influences of Asymmetric Heating on Hurricane Evolution in the MM5. J. Atmos. Sci.: 62, 3974–3992. Montgomery, M. T., and R. J. Kallenbach, 1997: A theory of vortex Rossby waves and its application to spiral bands and intensity changes in hurricanes. Quart. J. Roy. Meteor. Soc., 123, 435–465.

Montgomery, M. T., H. D. Snell, and Z. Yang, 2001: Axisymmetric Spindown Dynamics of Hurricane-like Vortices. J. Atmos. Sci., 58, 421-435. Montgomery, M. T., V. A. Vladimirov, and P. V. Denissenko, 2002: An experimental study on hurricane mesovortices. J. Fluid Mech., 471, 1–32.

Morrison, Ian, Steven Businger, Frank Marks, Peter Dodge and Joost A. Businger. 2005: An Observational Case for the Prevalence of Roll Vortices in the Hurricane Boundary Layer. J. Atmos. Sci.: 62, 2662–2673. Moss, M.S. and S.L. Rosenthal, 1975: On the estimation of planetary boundary layer variables in mature hurricanes. Mon. Wea. Rev., 103, 980-988.

Muramatsu, T., 1986: The structure of polygonal eye of a typhoon. J. Meteor. Soc. Japan, 64, 913–921.

National Research Council, 2004: Assessment of the Benefits of Extending the Tropical Rainfall Measuring Mission: A Perspective from the Research and Operations Communities, Interim Report. The National Academies Press, Washington, D. C. National Research Council, 2006: NOAA’s role in space-based global precipitation estimation and application. The National Academies Press, Washington, D.C.

114

Nolan, D.S. and B.F. Farrell, 1999a: Generalized Stability Analyses of Asymmetric Disturbances in One- and Two-Celled Vortices Maintained by Radial Inflow. J. Atmos. Sci., 56, 1282-1307. Nolan, D.S. and B.F. Farrell, 1999b: The intensification of two-dimensional swirling flows by stochastic asymmetric forcing. J. Atmos. Sci., 56, 3937-3962. Nolan, D.S., and Montgomery M.T., 2000: The algebraic growth of wavenumber one disturbances in hurricane-like vortices. J. Atmos. Sci. 57, 3514-3538. Nolan, D.S., M.T. Montgomery, and L.D. Grasso, 2001: The Wavenumber-One Instability and Trochoidal Motion of Hurricane-like Vortices. J. Atmos. Sci., 58, 3243-3270. Nolan, D.S. and M.T. Montgomery, 2002: Nonhydrostatic, Three-Dimensional Perturbations to Balanced, Hurricane-like Vortices. Part I: Linearized Formulation, Stability, and Evolution. J. Atmos. Sci.. 59, 2989-3020. Nolan, D.S., and L. D. Grasso. 2003: Nonhydrostatic, Three-Dimensional Perturbations to Balanced, Hurricane-Like Vortices. Part II: Symmetric Response and Nonlinear Simulations. J. Atmos. Sci.. 60, 2717-2745. Nolan, D.S., 2005: Instabilities in hurricane-like boundary layers. Dynamics of Atmospheres Oceans: 40, 209-236 Nolan, D.S., Y. Moon, and D. Stern, 2006: Tropical cyclone intensification from asymmetric convection: Energetics and efficiency. Submitted to J. Atmos. Sci. Nong, S. and K. Emanuel, 2003: A numerical study of the genesis of concentric eyewalls in hurricanes. Quart. J. Roy. Meteor. Soc., 129, 3323-3338. Nuissier, Olivier; R. F. Rogers; F. Roux. 2005 : A numerical simulation of Hurricane Bret on 22-23 August 1999 initialized with airborne Doppler radar and dropsonde data. Quart. J. Roy. Meteorol. Soc. 131, 155-194. Oda, Masahito, Toshihisa Itano, Gen’ichi Naito, Mikio Nakanishi and Kikuro Tomine. 2005: Destabilization of the Symmetric Vortex and Formation of the Elliptical Eye of Typhoon Herb. J. Atmos. Sci.: 62, 2965–2976. Ortt, D. and S.S. Chen, 2006: Rainbands and secondary eye wall formation as observed in RAINEX. Extended Abstracts, 27th Conference on Hurricanes and Tropical Meteorology, Amer. Meteor. Soc., Monterey, Ca, 24 - 28 April. CD-ROM. Paterson, Linda A., Barry N. Hanstrum, Noel E. Davidson and Harry C. Weber. 2005: Influence of Environmental Vertical Wind Shear on the Intensity of Hurricane-Strength Tropical Cyclones in the Australian Region. Mon. Wea. Rev.: 133, 3644–3660 Peng, M.S., B.-F. Jeng and R. T. Williams. 1999: A Numerical Study on Tropical Cyclone Intensification. Part I: Beta Effect and Mean Flow Effect. J. Atmos. Sci., 56, 1404–1423.

Persing, J., M.T. Montgomery and R.E. Tuleya, 2002: Environmental interactions in the GFDL hurricane model for Hurricane Opal. Mon. Weea. Rev., 130, 298 - 317.

Persing, J. and M. T. Montgomery. 2003: Hurricane Superintensity. J. Atmos. Sci.. 60, 2349-2371. Persing, J. and Michael T. Montgomery. 2005: Is Environmental CAPE Important in the Determination of Maximum Possible Hurricane Intensity?. J. Atmos. Sci.: 62, 542–550.

115

Powell, M.D., 1980: Evaluations of diagnostic marine boundary-layer models applied to hurricanes. Mon. Wea. Rev., 108, 757-766.

Powell, M.D., 1982: The transition of the Hurricane Frederic boundary-layer wind fields from the open Gulf of Mexico to landfall. Mon. Wea. Rev., 110, 1912-1932.

Powell, M.D., 1987: Changes in the low-level kinematic and thermodynamic structure of Hurricane Alicia (1983) at landfall. Mon. Wea. Rev., 115, 75-99.

Powell, M.D., 1990a: Boundary layer structure and dynamics in outer hurricane rainbands. Part I: Mesoscale rainfall and kinematic structure. Mon. Wea. Rev., 118, 891-917.

Powell, M.D., 1990b: Boundary layer structure and dynamics in outer hurricane rainbands. Part II: Downdraft modification and mixed layer recovery. Mon. Wea. Rev., 118, 918-938.

Powell, M.D. and P.G. Black, 1990: The relationship of hurricane reconnaissance flight-level wind measurements to winds measure by NOAA=s oceanic platforms. J. Wind Engineering And Industrial Aerodynamics, 36, 381-392.

Powell, M.D., P.D. Dodge and M.L. Black, 1991: The landfall of Hurricane Hugo in the Carolinas: Surface wind distribution. Wea. and Forecasting, 6, 379-399.

Powell, M.D., S.H. Houston and T.A. Reinhold, 1996: Hurricane Andrew=s landfall in south Florida. Part I: Standardizing measurements for documentation of surface wind fields. Wea. and Forecasting, 11, 304-328.

Powell, M.D. and S.H. Houston, 1996: Hurricane Andrew=s landfall in south Florida. Part II: Surface wind fields and potential real-time applications. Wea. and Forecasting, 11, 329-349.

Powell, M.D. and S.H. Houston, 1998: Surface wind fields of Hurricanes Erin, Opan, Luis, Marilyn and Roxanne at landfall. Mon. Wea. Rev., 126, 1259-1273.

Powell, M.D., P.J. Vickery and T.A. Reinhold, 2003: Reduced drag coefficient for high wind speeds in tropical cyclones. Nature, 422, 279-283.

Reasor P. D., Montgomery M. T., Marks F. D. Jr., Gamache J. F., 2000: Low-wavenumber structure and evolution of the hurricane inner core observed by airborne dual-doppler radar. Mon. Wea. Rev. 128, 1653-1680. Reasor, P.D., and M.T. Montgomery, 2001: Three-dimensional alignment and corotation of weak, TC-like vortices via linear vortex Rossby waves. J. Atmos. Sci., 58, 2306-2330. Reasor, P.D., M.T. Montgomery, and L.D. Grasso, 2004: A new look at the problem of tropical cyclones in vertical shear flow: Vortex resiliency. J. Atmos. Sci. 61, 3-22. Reisner, J. M., A. Mousseau, A. A. Wyszogrodzki and D. A. Knoll. 2005: An Implicitly Balanced Hurricane Model with Physics-Based Preconditioning. Mon. Wea. Rev.: 133, 1003–1022 Romine, G.S. and R.B. Wilhelmson. 2006: Finescale spiral band features within a numerical simulation of Hurricane Opal (1995). Mon. Wea. Rev., 134, 1121–1139. Roux Frank, Chane-Ming F., Lasserre-Bigorry A. and Nuissier O.. 2004: Structure and Evolution of Intense Tropical Cyclone Dina near La Réunion on 22 January 2002: GB-EVTD Analysis of Single Doppler Radar Observations. J. Atmos. Ocean. Tech. 21, 1501–1518.

116

Rosenthal, S.L., 1962: A theoretical analysis of the field of motion in the hurricane boundary layer. National Hurricane Research Project Report No 56, Washington D.C., 12pp.

Rozoff, C.M., W.H. Schubert, B.D. McNoldy and J.P. Kossin. 2006: Rapid filamentation zones in intense tropical cyclones. J. Atmos. Sci., 63, 325–340.

Schecter, D.A., M.T. Montgomery, and P.D. Reasor, 2002: A theory for the vertical alignment of a quasigeostrophic vortex. J. Atmos. Sci., 59, 150-168. Schneider R. and G.M. Barnes, 2005: Low-Level Kinematic, Thermodynamic, and Reflectivity Fields Associated with Hurricane Bonnie (1998) at Landfall. Mon. Wea. Rev., 133, 3243-3259.

Schroeder, J.L. and D.A. Smith, 2003: Hurricane Bonnie wind flow characteristics as determined from WEMITE. J. Wind Engineering and Industrial Aerodynamics, 91, 767B789.

Schubert, W.H., M.T. Montgomery, R.K. Taft, T.A. Guinn, S.R. Fulton, J.P. Kossin and J.P. Edwards, 1999: Polygonal eyewalls, asymmetric eye contraction, and potential vorticity mixing in hurricanes. J. Atmos. Sci., 56, 1197-1223.

Schwendike, J., 2005: The Boundary Layer Winds in Hurricanes Danielle (1998) and Isabel (2003). Diplomarbeit in Meteorology, Freie Universitat, Berlin, 255pp. Shapiro, L.J., 1983: The asymmetric boundary layer under a translating hurricane. J. Atmos. Sci., 40, 1984-1998.

Shapiro, L.J., and M.T. Montgomery, 1993: A three-dimensional balance theory for rapidly rotating vortices. J. Atmos. Sci., 50, 3322-3335. Shapiro, L.J., 2000: Potential vorticity asymmetries and tropical cyclone evolution in a three-layer model. J. Atmos. Sci., 57, 3645-3662. Shapiro, L.J. and J.D. Möller, 2003: Influence of Atmospheric Asymmetries on the Intensification of Hurricane Opal: Piecewise PV Inversion Diagnosis of a GFDL Model Forecast. Mon. Wea. Rev.. 131, 1637-1649. Shapiro, L.J. and J.D. Möller, 2005: Influence of Atmospheric Asymmetries on the Intensification of GFDL Model Forecast Hurricanes. Mon. Wea. Rev.: 133, 2860–2875 Shen, W., I. Ginis, and R.E. Tuleya, 2002: An investigation of land surface water on landfalling hurricanes. J. Atmos. Sci., 59, 789-802.

Simpson, R., 2003, Hurricane!: Coping with disaster, American Geophysical Union, Washington, D.C., 360 pp. Chapters by Gray, Rappaport and Simpson, Sheets, DeMaria and Gross, Frank, Myers and White, Pielke, Emanuel, Willoughby, Velden et al., Ritchie et al., Tyrell and Holland, Emmitt, Holland. Smith, R. K., W. Ulrich, and G. Sneddon, 2000: The dynamics of hurricane-like vortices in vertical shear flows. Quart. J. Roy. Meteor. Soc., 126, 2653-2670. Smith, R.K., 2003: A simple model of the hurricane boundary layer. Quart. J. Royal Meteorol. Soc., 128, 1007-1027. doi: 10.1256/qj.01.197

Smith, R. K., 2005: "Why must hurricanes have eyes?" Revisited. Weather: 131, 326-328 Smith, R.K., M. T. Montgomery, and H. Zhu, 2005: Buoyancy in tropical cyclone and other rapidly

117

rotating atmospheric vortices. Dynamics of Atmospheres and Oceans: 40, 189-208. Smith, R.K, 2006: Accurate determination of a balanced axisymmetric vortex in a compressible atmosphere. Tellus. 58A, 98-103. Spencer, R.W., H.M. Goodman, R.E. Hood, 1989: Precipitation retrieval over land and ocean with SSM/I: Identification and characteristics of the scattering signal. J. Atmos. Ocean. Technol., 6, 254-273. Stern, D.P. and S.D. Aberson, 2006: Extreme vertical winds measured by dropwindsondes in hurricanes. Extended Abstracts, 27th Conference on Hurricanes and Tropical Meteorology, Amer. Meteor. Soc., Monterey, Ca, 24 - 28 April. CD-ROM. Tonkin, H., Holland, G.J., Holbrook, N. and Henderson-Sellers, A. 2000: An evaluation of thermodynamic estimates of climatological maximum potential tropical cyclone intensity. Mon. Wea. Rev., 128, 746-762. Tuleya, R.E., M.A. Bender, and Y. Kurihara, 1984: Simulation study of the landfall of tropical cyclones, using a movable nested-mesh model. Mon. Wea. Rev., 112, 124-136.

Tuleya, R.E., 1994: Tropical storm development and decay: sensitivity to surface boundary conditions. Mon. Wea. Rev., 122, 291-304.

Velden, C. S. and J. D. Hawkins, 2002, The increasing role of weather satellites in tropical cyclone analysis and forecasting, 5th International Workshop on Tropical Cyclones (IWTC-V), Cairns, Australia, Dec. Velden, C., B. harper, F. Wells, J.L. Beven II, R. Zehr, T. Olader, M.Mayfield, C. Guard, M. Lander, R. Edson, L. Avila, A. Burton, M. Turk, A. Ikuchi, A. Christian, P. Caroffe and P. McCrone, 2006: The Dvorak tropical cyclone intensity estimation technique: A satellite-based method that has endured for over 30 years. Bull Amer. Meteorol. Soc., in press.

Wakimoto, R.M., and P.G. Black, 1994: Damage survey of Hurricane Andrew and its relationship to the eyewall. Bull. Amer. Meteor. Soc., 75, 189-200.

Walsh, K.J.E. 2002: Probability of cyclone strike on specific regions of the Queensland coast, given current cyclone position and movement. Australian Meteorological Magazine. 51, 69-75. Wang, Y., and G.J. Holland, 1995: On the interaction of tropical-cyclone-scale vortices. IV: Baroclinic vortices. Quart. J. Roy. Meteor. Soc., 121, 95-126. Wang, Y., and G.J. Holland, 1996a: Beta drift of baroclinic vortices. Part I: Adiabatic vortices. J. Atmos. Sci., 53, 411-427. Wang, Y., and G.J. Holland, 1996b: Beta drift of baroclinic vortices. Part II: Diabatic vortices. J. Atmos. Sci., 53, 3737-3756. Wang, Y., and G.J. Holland, 1996c: Tropical cyclone motion and evolution in vertical shear. J. Atmos. Sci., 53, 3313-3332. Wang, Y., 2001: An Explicit Simulation of Tropical Cyclones with a Triply Nested Movable Mesh Primitive Equation Model: TCM3. Part I: Model Description and Control Experiment. Mon. Wea. Rev., 129, 1370-1394.

118

Wang, Yuqing, 2002a: Vortex Rossby Waves in a Numerically Simulated Tropical Cyclone. Part I: Overall Structure, Potential Vorticity, and Kinetic Energy Budgets. J. Atmos. Sci.. 59, 1213-1238. Wang, Yuqing, 2002b: Vortex Rossby Waves in a Numerically Simulated Tropical Cyclone. Part II: The Role in Tropical Cyclone Structure and Intensity Changes. J. Atmos. Sci.. 59, 1239-1262. Wang, Y., 2002c: An explicit simulation of tropical cyclones with a triply nested movable mesh primitive equation model: TCM3. Part II: Some model refinements and sensitivity to cloud microphysics parameterization. Mon. Wea. Rev., 130, 3022–3036 Wang, Y., M.T. Montgomery, and B. Wang, 2004: How much vertical shear can a well-developed tropical cyclone resist? Preprints of the 26th Conference on Hurricanes and Tropical Meteorology, 3-7 May 2004, Miami, Florida, 100-101. Wang, Y. and C.-C. Wu. 2004: Current understanding of tropical cyclone structure and intensity changes - a review. Meteorol. and Atmos. Phys.: 87, 257 - 278. Wessel, J., J. Cornelius, R. W. Farley, A. Fote, J. Haferman, B. Gardiner, Y. Hong, D. B. Kunkee, G. Poe, S .D. Swadley, D. J. Tesmer, B. H. Thomas, E. Uliana, and D. Boucher, 2004: First observations from the DMSP SSMIS, 13th AMS Conference on Satellite Meteorology and Oceanography.

Willoughby, H.E., 1978: A possible mechanism for the formation of hurricane rainbands. J. Atmos. Sci., 35, 838-848.

Willoughby, H.E., J.A. Clos and M.G. Shoreibah. 1982: Concentric Eye Walls, Secondary Wind Maxima, and The Evolution of the Hurricane vortex. J. Atmos. Sci., 39, 395–411. Willoughby, H.E., F.D. Marks Jr. and R.J. Feinberg. 1984: Stationary and Moving Convective Bands in Hurricanes. J. Atmos. Sci., 41, 3189–3211. Willoughby, H.E. and M.E. Rahn, 2004: Parametric representation of the primary hurricane vortex. Part I: Observations and evaluation of the Holland (1980) model. Mon. Wea. Rev., 132, 3033 - 3048. Willoughby, H.E., R.W.R. Darling and M.E. Rahn, 2006: Parametric representation of the primary hurricane vortex. Part II: A new family of sectionally continuous profiles. Mon. Wea. Rev., 134, 1102-1120. Wong, M. L. M. and J. C. L. Chan, 2004: Tropical cyclone intensity in vertical wind shear. J. Atmos. Sci.: 61, 1859-1876. Wroe, D.R. and G.M. Barnes. 2003: Inflow Layer Energetics of Hurricane Bonnie (1998) near Landfall. Mon. Wea. Rev., 131, 1600–1612. Wu, C.-C. and K.A. Emanuel. 1993: Interaction of a Baroclinic Vortex with Background Shear: Application to Hurricane Movement. J. Atmos. Sci., 50, 62–76. Wu C.-C., 2001: Numerical simulation of Typhoon Gladys (1994) and its interaction with Taiwan terrain using GFDL hurricane model. Mon. Wea. Rev., 129, 1533–1549. Wu C.-C., Emanuel K.A., 1993: Interaction of a baroclinic vortex with background shear: Application to hurricane movement. J. Atmos. Sci., 50, 62–76. Wu C.-C., Cheng H.-J., 1999: An observational study of environmental influences on the intensity changes of typhoons Flo (1990) and Gene (1990). Mon. Wea. Rev., 127, 3003–3031.

119

Wu C.-C., Kuo Y.-H., 1999: Typhoons affecting Taiwan: Current understanding and future challenges. Bull. Amer. Meteor. Soc., 80, 67–80. Wu C.-C., Yen T.-H., Kuo Y.-H., Wang Y., 2002: Rainfall simulation associated with typhoon Herb (1996) near Taiwan. Part I: The topographic effect. Wea. Forecast., 17, 1001–1015. Wu C.-C., Chou K.-H., Cheng H.-J., Wang Y., 2003: Eyewall contraction, breakdown and reformation in a landfalling typhoon. Geophys. Res. Lett., 30, 1887; DOI: 10.1029=2003GL017653. Wu, Liguang and S.A. Braun. 2004: Effects of Environmentally Induced Asymmetries on Hurricane Intensity: A Numerical Study. J. Atmos. Sci.: 61, 3065-3081. Wurman, J. and J. Winslow, 1998: Intense sub-kilometre-scale boundary layer rolls observed in Hurricane Fran. Science, 280, 555-557.

Yau, M. K., Y. Liu, D.-L. Zhang and Y. Chen. 2004: A Multiscale Numerical Study of Hurricane Andrew (1992). Part VI: Small-Scale Inner-Core Structures and Wind Streaks. Mon. Wea. Rev.: 132, 1410-1433. Yeh, T.-C. and R.L. Elsberry. 1993: Interaction of Typhoons with the Taiwan Orography. Part I: Upstream Track Deflections. Mon. Wea. Rev., 121, 3193–3212. Yeh, T.-C. and R.L. Elsberry. 1993: Interaction of Typhoons with the Taiwan Orography. Part II: Continuous and Discontinuous Tracks across the Island. Mon. Wea. Rev., 121, 3213–3233. Zeng, Z., Y. Wang, and C.-C. Wu, 2006: Environmental dynamical control of tropical cyclone intensity - An observational study. Mon. Wea. Rev., in press. Zhang, D.-L., Y. Liu and M. K. Yau. 2000: A Multiscale Numerical Study of Hurricane Andrew (1992). Part III: Dynamically Induced Vertical Motion. Monthly Weather Review: 128, 3772–3788 Zhang, D.-L, Y. Liu, and M. K. Yau, 2001: A Multiscale Numerical Study of Hurricane Andrew (1992). Part IV: Unbalanced Flows. Mon. Wea. Rev., 129, 92-107. Zhang, D. L., Y. Liu, and M. K. Yau, 2002: A multiscale numerical study of hurricane Andrew(1992). Part V: Inner core thermodynamics. Mon. Wea. Rev., 130, 2745-2763. Zhang, Da-Lin; Kieu, Chanh Q. 2005: Shear-Forced Vertical Circulations in Tropical Cyclones. Geophys. Res. Lett.: 32, L13822. Zhang, Q.-H., S.-J. Chen, Y.-H. Kuo, K.-H. Lau and R.A. Anthes. 2005: Numerical Study of a Typhoon with a Large Eye: Model Simulation and Verification. Mon. Wea. Rev.: 133, 725–742 Zhu, Hongyan, Smith, Roger K., 2002: The Importance of Three Physical Processes in a Minimal Three-Dimensional Tropical Cyclone Model. J. Atmos. Sci.. 59, 1825-1840. Zhu, Hongyan, W. Ulrich and R. K. Smith. 2004: Ocean Effects on Tropical Cyclone Intensification and Inner-Core Asymmetries. J. Atmos. Sci.: 61, 1245-1258.

120

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES Topic 1.3 : Air-Sea Interface and Oceanic Influences Rapporteur: L. K. (Nick) Shay Center for Air-Sea Interaction Division of Meteorology and Physical Oceanography Rosenstiel School of Marine and Atmospheric Science University of Miami 4600 Rickenbacker Causeway Miami, FL 33149, USA Email: nshay@rsmas.miami.edu Fax: 305-421-4696 Working Group: P. G. Black, D. Barbary, M. Donelan, K. Emanuel, I. Ginis, C. Lozano, E. Uhlhorn, C.-C. Wu 1.3.1: Introduction The objective of this report is to document progress since IWTC-V in this thematic area from experimental, observational, empirical, theoretical, and numerical perspectives. The report begins by describing progress in upper-ocean processes that include the oceanic mixed layer (OML) and the thermocline. This section is followed by a discussion of the air-sea interface that includes surface winds and waves, and the communication to the atmospheric boundary layer through the momentum and enthalpy fluxes across the interface. These findings are summarized within a global context with specific recommendations on these important science issues to the WMO Commission on Atmospheric Science. 1.3.2: Upper-Ocean Processes Coupled oceanic and atmospheric models to predict hurricane intensity and structure change will eventually be used to issue forecasts to the public who increasingly rely on the most advanced weather forecasting systems to prepare for landfall (Marks and Shay 1998). For such models, it has become increasingly clear over the past decade that ocean models will have to include realistic initial conditions to simulate not only the oceanic response to hurricane forcing (Price 1981, 1994; Sanford et al. 1987; Shay 2001; D'Asaro 2003, Jacob and Shay 2003), but also to simulate the atmospheric response to oceanic forcing (Shay et al. 2000, 2006; Hong et al. 2000; Walker et al. 2005; Lin et al. 2005; Wu et al. 2006). An important example of this later effect was observed during the passages of hurricanes Katrina, Rita, and Wilma during the 2005 Atlantic Ocean hurricane season. Favorable atmospheric conditions prevailed in the Northwest Caribbean Sea and Gulf of Mexico (GOM) as the Loop Current (LC) extended several hundred kilometers north of the Yucatan Strait. As these storms moved over the deep warm pools, all three hurricanes explosively deepened and were more closely correlated with the ocean heat content (OHC) variations (and deep isotherms) than with the sea-surface temperatures (SST) distributions, which were essentially flat and exceeded 30oC over most of the region with slight warming along the northern GOM shelf. By contrast, the OHC and 26oC isotherm depths indicated the LC and its deep warm layers as it was in the process of shedding a mesoscale warm core ring (WCR) in August and September 2005. A cold core ring (CCR) that advected cyclonically around the shed WCR may have helped weaken Rita before landfall. Walker et al. (2005) also found that Hurricane Ivan

121

may have encountered a CCR prior to landfall. These studies point to the importance of initializing coupled ocean-hurricane models with realistic warm and cold ocean features.

Figure 1.3.1 a) Tropical cyclone image and b) schematic of the physical processes forced by hurricane winds such as shear-induced mixing and OML deepening, upwelling due to transport away from the center, and surface heat fluxes from the ocean to the atmosphere, all of which may contribute to ocean cooling during TC passage (from Shay 2001). As shown in Figure 1.3.1b, upper-ocean mixing and cooling are a strong function of forced near-inertial current shears that reduce the Richardson numbers below criticality, which induces entrainment mixing (Price 1981; Schade and Emanuel 1999; Shay 2001; D’Asaro 2003; Jacob and Shay 2003). The contributions to the heat and mass budgets by shear-driven entrainment heat fluxes across the ocean mixed layer (OML) base are 60 to 85%, surface heat fluxes are between 5 to 15%, and horizontal advection by ocean currents are 5 to 15 % (Price et al. 1994; Jacob et al. 2000). All of these processes impact the SST and OHC variability. In addition, wind-driven upwelling of the isotherms due to net upper ocean transport away from the storm modulate the shear-induced (S) ocean mixing events by an upward transport of cooler water from the thermocline. This transport increases the buoyancy frequency (N), which tends to keep the Richardson number above criticality. In the LC and WCR regimes with deep, warm thermal layers, cooling induced by these physical processes (Fig. 1) is considerably less as much greater turbulent-induced mixing by the current shear is required to cool and deepen the OML (Shay et al. 2000, Uhlhorn and Shay 2006). Quantifying the effects of forced current (and shear) on the OHC and SST distributions is central to accurately forecasting hurricane intensity and structure change.

122

Figure 1.3.2: OHC map and inset showing NRL mooring locations (red) and SRA wave measurements (black) relative to Ivan’s storm track and intensity. The OHC pattern shows the WCR encountered by Ivan prior to landfall. The cooler shelf water (OHC < 20 KJ cm-2) resulted from the passage of Frances two weeks earlier.

1.3.2.1 Oceanic Response

Hurricane Ivan (2-24 September 2004) moved over the NW Caribbean Sea with a radius of maximum wind (Rmax) of ~36 km (Fig. 1.3.2). Favorable environmental conditions of high OHC water (>150 KJ cm-2) plus outflow enhanced by upper-atmospheric flow ahead of an approaching trough helped Ivan maintain Cat-5 strength over 24-30 h (about one inertial period). Upon entering the GOM as a Cat-5 storm, Ivan weakened to a Cat-3 storm due to a combination of lower OHC, vertical shear in the atmosphere associated with an upper-level trough, and dry air being drawn into its circulation. Within 12 hours of landfall, Ivan encountered a WCR and surface pressure decreased by ~10 mb. However, cooler GOM shelf water forced by Hurricane Frances (10 days earlier) along with increasing shear both acted to oppose intensification during an eyewall replacement cycle. Thus, Ivan was an example of the impact of alternating positive and negative oceanic feedbacks on hurricane intensity. As shown in the Fig. 1.3.2 inset, the Naval Research Laboratory (NRL) had previously deployed several moorings about 180 km south of Mobile, Alabama as part of the Slope to Shelf Energetics and Exchange Dynamics (SEED) project from early May through early November 2004. Hurricane Ivan passed directly over these Acoustic Doppler Current Profiler (ADCP) moorings (Teague et al. 2005), which provided the temporal evolution of the 3-D current (and shear) structure at 2 to 4 m vertical resolution (Fig. 1.3.3). The current response at one of the Ivan moorings is shown for the near-inertial wave band (band-pass filter of the detided current signals) where the local inertial period is ~24 h. Profiler data starting from 50 m and extending to 492 m (4-m bins) were filtered between 22 to 28 hours. Second, the profiler data that were vertically averaged to estimate the depth-averaged current response (upper panels of Fig. 1.3.3) have similar amplitudes of 5 to 8 cm s-1 as those found in the limited vertical sampling from the Hurricane Frederic arrays (Shay and Chang 1997). These barotropic signals arrive before the storm owing to the large phase speeds of these oscillations.

123

Figure 1.3.3: Band-pass filtered cross-track (u) and along-track (v) time series at mooring 8 with t = 0 the time of hurricane passage and the time axis normalized by the inertial period. Shown top to bottom are depth averaged cross-track, baroclinic cross-track after the depth-average is removed (contoured), along-track depth- average, and along-track baroclinic (contoured). Time interval spans 15-29 September (Data courtesy NRL-Stennis). After the depth-averaged flows are removed, the band-pass filtered signals (lower panels in Fig.1.3.3) illustrate the ocean baroclinic response to hurricane forcing. Notice the downward propagation of the baroclinic energy from the wind-forced OML into the thermocline, which is consistent with modal separation (Gill 1984). The baroclinic motions have a characteristic time scale for the phase of each mode to separate from the wind-forced OML when the wind stress scale (2Rmax) exceeds the deformation radius associated with the first baroclinic mode (~ 30 to 40 km). This time scale increases with mode number due to decreasing phase speeds (Shay 2001). This vertical propagation is primarily associated with the predominance of the clockwise rotating energy (Shay and Jacob 2006). The vertical velocity signal in the upper ocean induced by Ivan was about 1.5 cm s-1. As shown in Fig. 1.3.4, similar strong near-inertial current response was observed by ElectroMagnetic Autonomous Profiling Explorer (EM-APEX) floats deployed in front of Hurricane Frances (2004) by Sanford et al. (2006). These Lagrangian profiling floats have provided a new view of near-inertial, internal wave radiation in unprecedented detail that includes not only the temperature and salinity (and thus density), but also the horizontal velocity structure. The phase propagation of the forced near-inertial currents is upward and is associated with downward energy propagation from the wind-forced OML. The shears (S) associated with these wind-forced currents lead to mixing events that significantly contribute to the observed ocean cooling of 2 to 2.5oC (and deepening) of the OML under the hurricane. Since the profilers also measure the density structure (and buoyancy frequency N), these Lagrangian floats provide the evolution of the Richardson numbers (N2 S-2) for the first time under

124

strongly forced hurricane conditions. Thus, the EM-APEX floats represent a new tool to improve our understanding of upper ocean processes and variability for a spectrum of conditions. For these near-inertial motions, the currents rotate clockwise with depth as wind-driven energy propagates downward into the thermocline while the phase propagates upward (Leaman 1976), which appears to be the case for the first few IP in the Ivan and Frances measurements. Rates of vertical energy flux forced by hurricanes have an average value of ~2 ergs cm-2 s-1 (Shay and Jacob 2006).

Figure 1.3.4: Current (U,V in m s-1), salinity (psu) and density or sigma-t (kg m-3) at Rmax during the passage of hurricane Frances (2004) as measured by the EM-APEX) floats developed by Webb and APL-University of Washington. Floats were deployed from a WC-130 by the USAF Reserve unit (from Sanford et al. 2006).

1.3.2.2 Ocean Heat Content Variability If the upper-ocean warm layer is thick and has a large total OHC, the SST will decrease slowly during TC passage, the negative feedback mechanism (or “brake”) will be weak, and the ocean will tend to promote TC intensification. Shay et al. (2000, 2006) showed that the OHC relative to the depth of the 26°C isotherm takes this into account and is a better indicator than just SST of the potential for TC intensification. Leipper and Volgenau (1972) estimated OHC as:

( )26

0

OHC 26 ,D

pc T z dzρ= − ∫

125

where cp is specific heat at constant pressure, D26 is the 26°C isotherm depth, and OHC is zero wherever water above 26°C is not present. OHC (and D26) is monitored using satellite remote sensing techniques (see http://iwave.rsmas.miami.edu/~nick/heat) using measurements from satellite-based radar altimeter estimates of the surface height anomaly (SHA) field from NASA TOPEX, Jason-1, U.S. Navy Geosat Follow-On-Mission (GFO) (Cheney et al. 1994; Scharroo et al. 2005) and SST cast within a two-layer model (e.g., Goni et al. 1996) and a seasonal climatology (Mainelli-Huber 2000). Since ocean features only move a few km d-1, altimeter-derived SHA locates warm (cold) mesoscale features that are usually identified as positive (negative) values as observed during Opal (Shay et al. 2000) and Ivan (Walker et al. 2005). These daily estimates are used in the Statistical Hurricane Intensity Prediction Scheme (SHIPS) to forecast intensity at the Tropical Prediction Center (DeMaria et al. 2005; Mainelli et al. 2006). Of relevance to the prediction problem, regions with thick (100-200 m) warm layers such as the Caribbean Sea, the LC, and WCR have high OHC (i.e., deep warm layers) and provide more sustained heat to the atmosphere during hurricane passage. Other regions such as the interior GOM, where the Gulf Common Water (GCW) has a smaller OHC distributed over a thinner (40 m) warm layer, tend to be less favorable for significant intensification (Jacob and Shay 2003). As shown in Fig. 1.3.5, both Katrina and Rita deepened to a Cat-5 hurricane over a lobe-like structure along the LC’s western flank as they moved at a speed of 5 to 6 m s-1. Notice the one-to-one correlation between hurricane intensity and OHC values of ~120 kJ cm-2 in the LC. The SSTs of more than 30oC were nearly uniformly distributed in this regime, and did not reveal the complex LC/WCR structure (Sun et al. 2006; Shay et al. 2006). Thus, the surface fluxes increased over the LC after Katrina emerged over the Gulf of Mexico where OHC values exceeded 100 kJ cm-2. This OHC level is more than five times the threshold of 16 kJ cm-2 d-1 integrated over the storm scale as suggested by (Leipper and Volgenau 1972). To compare along-track pressure fluctuations to along-track OHC and SST variations, longitude, latitude, and pressure (stars on pressure curve in Fig. 1.3.5b) from the best track 6-h data were used to estimate storm position and pressure at 2-hour intervals using linear interpolation. For both storms, the SST curve represents the along-track surface temperatures. Notice that normalized OHC values vary inversely to pressure changes (surface pressure decreases, OHC increases), whereas the normalized SST values are essentially flat during Katrina’s and Rita’s passage. Lin et al. (2005) analyzed remote sensing imagery in the western Pacific Ocean prior and subsequent to the passage of Maemi in 2003 (Fig. 1.3.6). They found that Typhoon Maemi’s intensity increased by 36 m s-1 over a eddy-rich regime. Using results in a Coupled Hurricane Intensity Prediction System (Emanuel 2003), the WCRs acted as an insulator between typhoons and the deeper, cooler thermocline water (Wu et al. 2006). That is, the SST response is significantly less as the OML is much deeper in these regimes similar to findings in the western Atlantic Ocean basin. Without initializing the model with a warm ocean feature, the simulated typhoon intensity was one category below the observed intensity. Lin et al. (2005) point out the importance of the eddy-rich regime associated with the western boundary current or Kuroshio during the passage of typhoons over these oceanic features. In the western Pacific Ocean, the Kuroshio plays the same role as the Gulf Stream in the western Atlantic Ocean basin (i.e., poleward advection of warm tropical water).

126

Figure 1.3.5: Left panels (a,c) represent pre-storm OHC (kJ cm-2:color) and 26oC isotherm depth (m: black contour) based on a hurricane season climatology, Reynolds SSTs, Jason-1 and GFO radar altimetry measurements relative to the track and intensity of Hurricane a) Katrina and c) Rita. Right panels (b,d) represent time series of surface pressure (thin black) versus along-track SST (dashed) and OHC (thick black) variations normalized by 30oC and 60 kJ cm-2 , respectively. OHC uncertainty limits are based on 6-hourly values averaged in the cross-track direction between +/- 0.5o lat. from the track (from Shay et al. 2006). On 15 and 26 September 2005, oceanic current, temperature and salinity measurements were acquired from Airborne eXpendable Current Profilers (AXCP), Airborne eXpendable Conductivity Temperature and Depth (AXCTDs) profilers, and Airborne eXpendable Bathythermographs (AXBTs) in a pattern centered on the LC/WCR (Rogers et al. 2006; Shay et al. 2006). The 15 September flight was originally conceived as a post-Katrina experiment in an area where the hurricane rapidly intensified over the LC/WCR complex.

127

Figure 1.3.6: A composite of TMI and AMSR-E passes during 3-5 September 2003 showing pre-Maemi SST distribution relative to Maemi’s track over the locations of WCR-rich zones (From Fig. 2 of Lin et al. 2005). To obtain in-situ measurements of OHC within the same WCR, two thermistor chain drifters were deployed from the USAF WC-130 aircraft in the path of Hurricane Rita on 21 September 2005. As shown in Fig. 1.3.7, upper ocean T(z) time series were obtained from two drifters that were closest by Rita (distance in upper panels) while circulating around the WCR. Notice that these two drifters were about 80 to 100 km from the WCR center with OHC values of 120 kJ cm-2 where the 26oC isotherm depths were between 90 to 100 m. Altimeter-derived OHC values in the WCR ranged from 105-120 kJ cm-2 so these altimeter values are consistent with the low-pass filtered values in the right panels prior to Rita. The estimated enthalpy fluxes exceeded 2000 W m-2 from one of the drifters (not shown) that passed closest to Rita while it was at Category 4 strength with Rmax < 20 km. Rita then clipped the northeastern part of the WCR as the storm was weakening prior to landfall on the Texas-Louisiana border as a Cat-3 hurricane. As during Hurricane Opal (Shay et al. 2000; Hong et al. 2000), these in situ measurements and inferred surface enthalpy fluxes are invaluable to help the TC communities understand the atmospheric response to ocean forcing.

128

Figure 1.3.7: Thermal structure (oC: left panels) and observed and low-pass filtered (solid black) of OHC (kJ cm-2: right panels) time series from two drifters deployed prior to Rita (black curve) relative to the WCR center (red) as depicted in the upper panels for the thermal structure in September 2005. The point of closest approach is shown as the dotted line (figure courtesy of Rick Lumpkin of NOAA/AOML). Using these in situ profiles combined with radar-altimetry fields, isotherm depths and OHC values are compared to assess uncertainties in satellite retrievals for pre- and post-Rita cases (Fig. 1.3.8). Satellite-inferred and in situ structures are well correlated (0.9) for both isotherm depths and OHC variations using Reynolds SSTs (Shay et al. 2006). Regression slope for the OHC is 0.9 with a bias of 1.3 kJ cm-2 by combining pre and post-Rita data set in the WCR only. For the 26oC isotherm depths, the slope was about 1.1 with a 9.3 m bias where the altimeter-derived value was larger than that from the profiler data. This larger bias was primarily associated with the advection of the CCR between the LC and WCR from the post-Rita data set on 26 Sept. These estimates were also consistent with those derived from drifter-based measurements. While the bias in the depth is quite large, the result suggests this is roughly a 10 to 15% uncertainty in the signals where isotherm depths ranged from 90 to 105 m in the WCR. Comparisons of several sets of profiler measurements have suggested that the OHC scales as ~0.9 to 1 kJ cm-2 m-1 in the LC and WCR structures, which suggests that if the 26oC isotherm depth is known, the OHC value scales with this empirical result.

129

Figure 1.3.8: Regression analysis of a) OHC and b) 26oC isotherm depth (m) for pre-Rita (black circles) and post-Rita (red circles) between in situ (abscissa) and satellite-inferred (ordinate) for the data acquired in the WCR/CCR regime on 15 and 26 September 2005.

1.3.2.3 Background Ocean States Background oceanic flows that are set up by large horizontal pressure gradients due to T(z) and S(z) may play a significant role in altering the development of strong wind-driven current shears within the LC/WCR complex as suggested in Fig. 1.3.9 (Shay and Uhlhorn 2006). Pre- and post-Isidore measurements across the Yucatan Strait indicate strong density and pressure gradients that are associated with the northward-flowing LC at speeds of up to 1 m s-1. In the post-Isidore case, the horizontal gradients were sharpened since the storm cooled the Yucatan shelf waters by more than 4oC compared to less than 1oC across the Yucatan Strait. Here, strong horizontal advection of the thermal and salinity gradients through this regime impacted the oceanic response within the LC. Falkovich et al. (2005) introduced an approach for feature-based ocean modeling that involves cross-frontal “sharpening” of the background temperature and salinity fields according to data obtained in field experiments, which allows specifying the position of the LC in the GOM using available observations. Briefly, the LC is a highly variable ocean feature in time as it can penetrate ~500 km northward of the Yucatan Strait. Recurring WCR shedding events with peak periods from 6 to 11 months (Sturges and Leben 2000) occur when CCRs are located on the LC periphery prior to separation. These WCRs, with diameters of ~200 km, propagate west to southwest at phase speeds of 3 to 5 km d-1 (Elliot 1982), and can remain in the Gulf of Mexico for several months. In this LC regime, OHC values relative to the 26oC isotherm exceed 100 kJ cm-2 (Leipper and Volgenau 1972). Such OHC levels are resistive to storm-induced cooling by wind-driven current shears across the OML base. Nof and Pichevin (2001) and Nof (2005) suggest that the LC cycle can be explained in terms of the momentum imbalance paradox theory. When a northward-propagating anomalous density current (Yucatan Current) flows into an open basin (GOM) with a coast on its right (Cuba), the outflow expands near its source to form a clockwise-rotating bulge (LC). The expansion of the current (WCR formation) is a necessary condition to satisfy the momentum flux balance along the northern coast of Cuba. Two-thirds of the outflow mass flux goes into this expanding bulge with the remaining flux forcing the downstream current. The subsequent WCR separation is due either to planetary vorticity gradients or topographic

130

effects (Cherubin et al. 2005), where ~80% of the inflow forces the current and the remaining inflow goes into a WCR (Nof 2005). This theory predicts a mass partition that has been observed during WCR shedding events in North Atlantic Ocean simulations.

Figure 1.3.9: Pre-(upper) and post-Isidore (lower) thermal (oC: color) and northward (into the page) geostrophic velocity (m s-1: dashed) cross-section from expendables deployed on 19 and 23 September 2002 across the Yucatan Straits. Heavy dashed line represents the 26oC isotherm depth (from Shay and Uhlhorn 2006). Data assimilative ocean nowcasts are an effective method for providing initial and boundary conditions to the ocean component of nested, coupled TC prediction models. The thermal energy available to intensify and maintain a TC depends on both the temperature and thickness of the upper ocean warm layer. The ocean model must be initialized so that ocean features associated with relatively large or small ocean heat content (OHC) are in the correct locations and T-S (and density) profiles, along with the OHC, are realistic. Ocean nowcast-forecast systems based on HYCOM (Bleck 2002; Chassignet et al. 2003; Halliwell 2004) were evaluated in the northwest Caribbean and eastern Gulf of Mexico for September 2002 prior to Hurricanes Isidore and Lili, and for September 2004 prior to Hurricane Ivan. In this region, the OHC distribution associated with the LC/WCR complex as well as coastal upwelling must be accurately initialized for the ocean model. In this context, measurements are critically important not only for assimilation, but to evaluate initial and boundary ocean fields.

131

Figure 1.3.10: OHC (kJ cm-2) in the northwest Caribbean Sea and southeast GOM from an objective analysis of aircraft observations, satellite altimetry, HYCOM NRL-CH nowcast, and HYCOM NRL-MODAS nowcast (four left panels). Temperature (right top) and salinity (right bottom) vertical profiles at a location in the northwest Caribbean Sea, where red lines are climatological profiles (GDEM3 dashed, WOA01 solid), solid blue lines are observed profiles, dashed blue lines are MODAS profiles, and black lines are model nowcasts (HYCOM-NRL dashed and HYCOM-MODAS solid). An examination of the initial analysis prior to Isidore is from the experimental HYCOM nowcast-forecast system of the NRL experiments in the Atlantic basin at 0.08° resolution. This model product assimilates both satellite altimetry SHAs (Cooper and Haines 1996) and optimally interpolated SSTs. Comparison of OHC maps hindcast by HYCOM to OHC maps objectively analyzed from aircraft measurements and derived from satellite observations (left panels of Fig. 1.3.10) demonstrate that this HYCOM analysis (labeled HYCOM NRL-CH in the figure) reproduces the LC orientation but underestimates OHC by ~50%. In the NW Caribbean Sea, the T(z) hindcast tends to follow the September climatology but does not reproduce the larger OHC values. In the HYCOM hindcast, the upper ocean is less saline than both climatology and observations above 250 m (Fig. 1.3. 10) and less saline than the observations between 250 and 500 m. HYCOM structure was subsequently relaxed to the Navy three-dimensional MODAS (Fox et al. 2002) and T-S analyses were generated from all available in-situ observations. The first HYCOM NRL-CH nowcast was adversely impacted by a poor initialization that could not be corrected by including only the SHA fields. Biases were reduced in this HYCOM NRL-MODAS product compared to observations in both horizontal maps and vertical profiles. Evaluation of the next-generation NRL nowcast-forecast system (Cummings 2003) is being done by performing hindcasts from mid-2003 to the present for Ivan (2004) and Katrina (2005) and Rita (2005). Initial evaluation of pre-Ivan conditions is encouraging because the large cold bias was no longer present, and because the LC/WCR complex (includes the CCR) was well represented (see Figs. 1.3.2 and 1.3.6). Such evaluations of model-generated products are needed prior to coupling with a hurricane model to insure that ocean features are in the correct place and have structural characteristics that are realistic.

132

1.3.2.4 Vertical Mixing Parameterization One of the significant effects on the upper-ocean heat budget and the fluxes to the atmosphere is the choice of entrainment mixing parameterizations. Jacob et al. (2006) have conducted sensitivity tests using five schemes: K-Profile Parameterization (KPP: Large et al. 1994); Goddard Institute for Space Studies Level-2 closure (GISS: Canuto et al. 2002); Level-2.5 turbulence closure scheme (MY: Mellor and Yamada, 1982); quasi-slab dynamical instability model (PWP: Price et al. 1986); and the turbulent balance model of Gaspar (KT: 1988) that is a modified version of Kraus and Turner (1967). As shown in Fig. 1.3.11 for quiescent ocean initial conditions, the range of fluxes in the directly forced region of Hurricane Gilbert exceeds 500 Wm-2 for these five schemes. For the Hurricane Gilbert case, Jacob and Shay (2003) simulated OML temperatures for realistic initial conditions and compared with profile observations to identify appropriate mixing schemes. The three higher-order turbulent mixing schemes (KPP, MY, GISS) considered will lead to a more accurate ocean response simulation. However, these comparisons are limited by data availability and therefore routine measurements are necessary to evaluate the ocean component of the coupled system. Similar to the post-season track and intensity verifications, more ocean observations must be acquired to evaluate the different schemes to build a larger statistical base. Given the large range in the simulated surface fluxes for different schemes, this is a crucial step toward reducing this uncertainty. The approach of stand-alone ocean simulations using derived realistic atmospheric forcing used here allows us to evaluate the ocean model and associated parameterizations. Since boundary layer forcing structure from the atmospheric component of the coupled model is subject to additional uncertainties, this approach will eventually lead to reduction in uncertainties of the ocean component in the coupled system based on observations. The OML salinity evolution with and without precipitation forcing highlights the effect of precipitation on the upper ocean salt budget (not shown). For the no precipitation case, OML salinity variability simulated by all the above mixing schemes was similar ahead of the storm center. In the PWP scheme, however, salinity increases significantly in the right-rear quadrant over the first half of an inertial period due to due to enhanced mixing. While

133

Figure 1.3.11: Evolution of OML quantities at a location 2 Rmax right of the storm track for quiescent ocean initial conditions in the Gilbert case (Jacob and Shay 2003): a) Mixed layer temperature (° C), b) Mixed layer salinity (PSU), and c) Fluxes to the atmosphere. Note the wide range of variability in the OML temperature and in the fluxes to the atmosphere for the five mixing parameterizations (see inset in panel (c)). in the other mixing cases, the simulated salinity evolution is similar with minimal changes in the KT model due to less intense mixing (no vertical shear). By including precipitation forcing, the salinity in the OML began to decrease about 0.5 IP before the storm with maximum freshening of the OML observed in the KT case. This freshening process due to enhanced precipitation increases static stability in the mixed layer, leading to a simulated salinity balance for PWP case that is more consistent with the other schemes. OML temperature and salinity evolution in cases with and without precipitation for PWP scheme indicates a mean temperature and salinity differences of 0.5oC and 0.2 psu in the OML. layer. An average freshening of 0.2 psu is seen in the wake of the storm in all the five cases when precipitation forcing was used, which is consistent with CTD measurements acquired during the Spectrum 90 expeditions (Pudov and Petrichenko 2000). Precipitation temperatures have minimal effect on the salinity evolution in the OML. Simulated results from the three-higher order schemes did not differ significantly from each other. 1.3.3: Air-Sea Interface: Because the underlying ocean significantly affects TC intensity, attention has been drawn toward gaining a better understanding of the physical interaction between the atmosphere and ocean during these events. Unfortunately, due to limited observations at the air-sea interface in high-wind conditions, the understanding has not progressed nearly enough to significantly improve the parameterization of momentum and energy transfer. The relationships of the transfer processes to small-scale roughness (Charnock 1955) and surface-layer stability (Monin-Obukhov similarity theory) are fairly well understood under low- to moderate-wind conditions (Large and Pond 1981), but additional phenomena not typically observed such as the maturity of the sea state (Donelan 1990) and sea spray (Fairall et al. 1994; Wang et al. 2001) have also been shown to modulate the heat and momentum exchange. These effects under TC-force winds have been primarily studied in controlled laboratory experiments

134

(Donelan et al. 2004). In a TC environment, both young and mature waves are present and impact the air-sea fluxes and OML and ABL processes. The TC intensity is maintained in part by the balance between the heat gained by the boundary layer of the storm and the energy lost due to friction. Emanuel (l986) proposed a theory requiring this relative balance to be the primary modulator of intensity. Based on this view, it is assumed that under certain conditions there should be a level of mutual dependence of the air-sea transfer processes of heat and momentum. Indeed, it has been suggested through highly idealized model simulations (Ooyama 1969; Rosenthal 1971; Emanuel 1995; Braun and Tao 2000) that the TC intensity is sensitive to the ratio of enthalpy transfer coefficient to drag coefficient (Ck Cd

-1). The conclusion that this quantity probably lies within a rather limited range (<1.5), is consistent with the observation that most TCs do not reach their maximum potential intensity (DeMaria and Kaplan 1993).

1.3.3.1 Surface Wave Field On moored NOAA buoys, wave spectral energy is derived from accelerometers or inclinometers that measure the heave accelerations, or vertical displacements, of moored pitch-roll buoys that use the new Multi-functional Acquisition and Reporting System (MARS) payload system. When Hurricane Emily approached buoy 44014 from the southeast with sustained winds of 28 m s-1, the significant wave heights reached 8 m (Fig. 1.3.12a). Maximum wave spectral energies were largely contained in the swell portion of the spectrum (i.e., ~13 s wave) that decayed rapidly after Emily’s passage. These wave spectral energies contained in the lower-frequency intervals associated with the swell began to increase several hours in advance of storm passage, peaked at the point of closest approach, and subsequently decayed over 1-2 inertial periods (IP~20 h) in the wake. In frequency bands between 0.2-0.4 Hz, the modulation of the wave spectral energies continued for an extended period of time (Faber et al. 1997). Wave spectral energies and significant wave heights indicate peaks that occur over IP time scales. More recently, Hurricane Lili approached buoy 42001 from a south-southwest direction after crossing the western tip of Cuba, just as she reached her maximum intensity (Cat- 4). As Lili passed within ~ Rmax to the west of the buoy, winds reached 48 m s-1, significant wave heights peaked above 10 m, and wave spectral energies exceeded 220 m2 Hz-1 (Fig. 1.3.12b). This is the region of the storm where the maximum ocean response is observed compared to the other quadrants. The largest values of wave spectral energy are again concentrated in the lower frequency band with maximum values at 0.4 IP (~10 h) prior to passage and persist for about 1.5 IP (~40 h) after the closest approach of Lili. Smaller amplitudes of the wave spectral energy in the higher frequency (0.2-0.4 Hz) intervals are evident 1.7 IP (~48 h) prior to passage, and persist after Lili’s passage over the buoy. The phase of the oscillations, which is most pronounced between the frequency intervals 0.2-0.4 Hz, is near the local IP (~27 h). Such observations should be used to test coupled ocean-wave models to assess their performance under strong wind and wave conditions.

135

Figure 1.3.12 : Time series of a) wind speed (solid) and significant wave height (dotted),and b) contoured wave spectral energy (0.1 to 1 m2 Hz-1) from buoys 44014 and 42001 during and subsequent to tropical cyclones Emily (left panels) and Lili (right panels). Time series are scaled by local inertial periods of ~20 hrs for buoy 44014 and ~27 hrs for 42001. Wang et al. (2005) documented the wave response to Ivan over the NRL SEED moorings. Wave heights significantly increased with peak values when the radial distance between the mooring and storm center was ~75 km (Fig. 1.3.13c). Hs reached maximum values of 16 m to 18 m on moorings 3, 4, and 5 and were larger than those observed at the NDBC buoy (15.9 m). The maximum wave height was recorded to be 27.7 m at mooring 3, and wave height variations were consistent with the radial variations in the surface wind of Ivan. At Rmax, the model predicted a maximum wave height of ~21 m (Fig. 1.3.13c). Previous studies have suggested in a hurricane wave field that the maximum wave height approaches 1.9 x Hs, which would be consistent with these measurements. The moored measurements sampled only a small part of the domain influenced by Ivan’s broad wind field. SRA measurements (Wright et al. 2001) from the aircraft should be used to extend the mooring measurements to a scale commensurate with Ivan’s wind field as suggested in Fig. 1.3.2.

136

Figure 1.3.13: a) Hurricane Ivan satellite image at 1850 UTC 15 September 2004 in the GOM with the green line representing the track of Ivan at 3-h intervals moving over the SEED moorings (blue). Panel b) represents the evolution of Hs (circles) and Hmax (crosses) as a function of normalized distance relative to Rmax. Hs is from NDBC buoy 42040 (dotted curve) and its radial distance to Ivan’s center is shown by the green squares. Panel c) represents Hs and Hmax as a function of normalized distance from the center compared to the exponential distance: digitized values of a segment 15o CW from the forward direction of a numerically simulated wave field (black asterisks). Blue curve depicts the line of Hmax=1.9Hs where circles and crosses are as in panel b) (from Wang et al. 2005).

1.3.3.2 Surface Winds

Surface winds in hurricanes have been estimated remotely using the Stepped-Frequency Microwave Radiometer (SFMR) from aircraft (Uhlhorn and Black 2003). With six frequencies, the SFMR measures radiative emissions, expressed in terms of brightness temperatures (Tb), from the ocean and the atmosphere. The percentage of foam coverage on the sea surface is known to increase monotonically with wind speed. At microwave frequencies, foam acts as a blackbody emitter. As foam increases, the ocean emits microwave energy more readily, and assuming a constant SST, the Tb increases. Given an accurate physical model that relates ocean surface wind speed and rain-rate to measurements of Tb at frequencies, a set of equations may, in theory, be inverted to calculate the surface winds. Uhlhorn et

137

al. (2006) has assessed SFMR measurements on aircraft for operational surface wind speed measurements in the active 2005 hurricane season, which provided ample data to evaluate both instrument performance and surface wind speed retrieval quality up to Cat-5 hurricanes. A new microwave emissivity and wind speed model function based on comparisons with direct measurements of surface winds in hurricanes by GPS dropwindsondes is shown in Fig. 1.3.14. This function eliminates a previously-documented high bias in moderate SFMR-measured wind speeds (10-50 m s-1), and additionally corrects an extreme wind speed (>60 m s-1) systematic underestimate in the past cases. The model function behaves differently below and above the hurricane wind speed threshold (32 m s-1), which may have implications for air-sea momentum and kinetic energy exchange. The change in behavior is at least qualitatively consistent with recent laboratory and field results concerning the drag coefficient (cd) in high wind speed conditions, which show a fairly clear ``leveling-off'' of cd with increased wind speed above ~30 m s-1 as discussed below.

1.3.3.3 Surface Drag Coefficients Since the energy source for the TC is the ocean, knowledge of the heat and moisture fluxes across the interface and into the ABL are critical elements. However, the exchange coefficients for heat, moisture, and momentum are not well known for the high wind speed and ocean surface wave conditions. Momentum transfer between the two fluids is characterized by the variations of wind with height and a surface drag coefficient that is a function of wind speed and surface roughness. However, it is difficult to acquire flux measurements for the high wind and wave conditions under the eyewall. Since 1997, GPS sondes have been deployed from aircraft to measure the Lagrangian wind profiles in the atmospheric boundary layer in TCs. Powell et al. (2003) found a logarithmic variation of mean wind speed in the lowest 200 m, a maximum speed at 500 m, and a gradual weakening with height to 3 km. From these estimates, the surface stress, roughness length, and neutral stability drag coefficient determined by the profile method suggest a leveling of the surface momentum flux as winds increase above hurricane-force and a slight decrease of the drag coefficient with increasing winds.

Figure 1.3.14: Excess emissivity from SFMR compared to 10-m (surface) winds measured from GPS dropsondes during the 2005 season. The total number of samples is 160 and the rms difference between the SFMR model function was 0.011 (from Uhlhorn et al. 2006).

138

Donelan et al. (2004) describe a series of wind-wave tank experiments that included stress measurements from hot-film anemometry and digital particle image velocimetry, and laser/line scan cameras for measuring the water surface elevation. Reynolds stress was measured directly with an x-film anemometer at low and moderate (centerline) wind speeds (0 to 26 m s-1). The stress determined at the measured elevations was used to correct the values at the surface with the measured horizontal pressure gradient in the tank. At higher winds, surface stress was determined from a momentum budget of sections of the tank. The steady-state surface stress increases the momentum of the wave field with increasing fetch, which drives a downwind current near the surface and maintains a downwind slope of the mean surface (mean surface elevation increasing in the downwind direction). The horizontal pressure gradient drives a return flow (upwind flow) in the bottom of the water column, which causes a drag on the bottom of the tank. Finally, the horizontal pressure gradient in the air that produces the wind adds to the slope of the water surface - the "inverted barometer" effect. Measurements of the drag coefficient from this laboratory experiment, referenced to the 10-m wind speed, are shown in Fig. 1.3.15. Wind speed was measured at 30 cm height in the tank and extrapolated to the 10-m using the logarithmic dependence on height and was verified between crest height and 30 cm for all but the two highest wind speeds. Three other data sets were obtained in the wind-wave tank using the profile method (in which the vertical gradient of mean horizontal velocity is related to the surface stress), the Reynolds stress method, and the momentum budget or “surface slope” method. The agreement among the various methods validates the momentum budget method which, being insensitive to airborne droplets, allows a measurement of the surface stress at the highest winds generated. Notice the characteristic behaviour of the drag coefficient as the surface conditions changes from aerodynamically smooth (characterized by a decrease in the drag coefficient at low-wind speeds) to aerodynamically rough (drag coefficient increasing with wind speed) conditions. In rough flow, the drag coefficient is related to height of the “roughness elements” per unit distance downwind or, more precisely, the spatial average of the downwind slopes. Unlike a solid surface, the roughness elements (or waves) are responsive to the wind so that the drag coefficient increases between 3 and 33 m s-1.

139

0 10 20 30 40 50 600

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5x 10

−3

Wind speed extrapolated to height of 10 m, U10

[m/s]

Drag

Coe

ff. re

ferre

d to 1

0 m

Laboratory measured drag coefficients by various methods

Squares = profile method (Ocampo−Torres et al., 1994)

Asterisks = profile method (This paper)

Circles = momentum budget (This paper)

Diamonds = Reynolds stress (This paper)

Dots = dissipation (Large and Pond, 1981)

Figure 1.3.15: Laboratory measurements of the neutral stability drag coefficient (x 10-3) by profile, eddy correlation (“Reynolds”), and momentum budget methods. The drag coefficient refers to the wind speed measured at the standard anemometer height of 10 m. The drag coefficient formula of Large and Pond (1981) is also shown along with values from Ocampo-Torres et al (1994) derived from field measurements (from Fig. 2 in Donelan et al. 2004). In a TC, the wind changes direction and speed over relatively short distances compared to those required to approach full wave development. Consequently, the largest waves in the wind-sea move relatively slowly compared to the wind and often travel in directions differing from the wind. Under such circumstances, these long waves contribute to the aerodynamic roughness of the sea as hypothesized by Kitaigorodskii (1968) and demonstrated by Donelan (1990). Measurements at sea (e.g., Large and Pond, 1981) and in laboratories (e.g., Donelan 1990; Ocampo-Torres et al. 1994) amply demonstrate the increasing aerodynamic roughness with increasing wind speed. A “saturation” of the drag coefficient does appear once the wind speed exceeds 33 m s-1 (Fig. 1.3.15). Beyond this speed, the ocean surface simply does not become any rougher in an aerodynamic sense. At the highest wind speed, the significant height and peak frequency of the waves in the laboratory were 9 cm and 1.4 Hz. In the range of wind speeds of 10 to 26 m s-1, the laboratory measurements parallel the open-ocean measurements (Large and Pond 1981), but are lower. These measurements suggest aerodynamic roughness saturation beyond 10 m height wind speeds of 33 m s-1. The saturation level for the drag coefficient is 0.0025, which corresponds to a roughness length of 3.35 mm. Powell et al. (2003) show “saturation” of the drag coefficient at 0.0026 at about 35 m s-1. Shay and Jacob (2006) also find a “saturation” at about 30 m s-1, but their analytical function, derived by equating internal wave fluxes to the surface stress, revealed a value 0.0034 before this tapering-off trend. The possibility of a limiting state in the aerodynamic roughness of the sea surface is of critical importance in understanding and modelling the development of hurricanes and other intense storms. Donelan et al. suggest a change in flow characteristics leading to saturated aero-dynamic roughness at boundary layer wind speeds in excess of 33 m s-1. Obviously, more research is needed to quantify the surface drag coefficient given its importance for calculating the enthalpy fluxes (Emanuel 2003).

140

1.3.3.4 Wind-Wave Coupling

Figure 1.3.16: Drag coefficients (Cd) from various observation-based values, empirical formulas, and model outputs as a function of U10. Symbols represent observations from GPS sonde wind profiles (Powell et al. 2003). Vertical bars represent 95% confidence limits. Solid line is an extrapolation of the Large and Pond (1981) formula. Dash-dot line is the bulk formula used in GFDL hurricane model. Shaded and hatched areas represent ranges between upper and lower bound of Cd obtained by the URI coupled wave-wind model and an internal estimation of WAVEWATCH III, respectively (from Moon et al. 2004). In the GFDL model, the momentum flux is parameterized with a constant non-dimensional surface roughness (or Charnock coefficient) and the stability correction is based on the Monin-Obukhov similarity theory, regardless of the wind speed or the sea state. In addition to wind speed, Donelan (1990) found Cd also depends on the sea state represented by the wave age. Moon et al. (2004) investigated the Charnock coefficient under TC conditions using a coupled wind-wave (CWW) model. In the CWW model, the surface wave directional frequency spectrum near the spectral peak is calculated using the WAVEWATCH III (Tolman 2002) model and the high frequency part of the spectrum was parameterized using the theoretical model of Hara and Belcher (2002). The wave spectrum is then introduced to the wave boundary layer model of Hara and Belcher (2004) to estimate the Charnock coefficient at differing wave evolution stages. They found that Cd (Fig. 1.3.16) levels off at high wind speeds, which is consistent with the above studies cited in Fig. 1.3.14. The most important finding of this study is that the relationship between the Charnock coefficient and the input wave age (wave age determined by the peak frequency of wind energy input) is not unique, but strongly depends on wind speed. The regression lines between the input wave age and the Charnock coefficient have a negative slope at low wind speeds but have a positive slope at high wind speeds. This behavior of the Charnock coefficient provides one explanation why Cd under a TC, where seas tend to be “young,” is reduced in high wind speeds. In the presence of surface waves, the momentum flux from the atmosphere may be different from the flux into ocean if the wave field is spatially/temporally varying. That is, the momentum transfer from the atmosphere consists of two components: momentum transferred directly into the ocean current (by

141

viscosity) and momentum transferred to the waves (through the wave-induced stress). The total momentum transfer into ocean then consists of two components: momentum transferred directly from the atmosphere (by viscosity) and momentum transferred from the waves (through the wave breaking). The difference between the fluxes into and out of the waves may be particularly significant under hurricane conditions since the wave field is far from fully-developed. This difference can reach up to 20% in the vicinity of Rmax, which may potentially have an important impact in the coupled hurricane-wave-ocean model.

1.3.3.5 Enthalpy Fluxes Fluxes of heat and moisture are central to the TC intensity question and are usually determined from bulk formulae that utilize near-surface atmospheric thermodynamic and wind observations and upper-ocean temperature data measured by expendable probes. Specifically, the atmospheric variables are estimated from the large number of GPS sondes (Hock and Franklin 1998) deployed within the storm from both NOAA aircraft as well as Air Force Reserve reconnaissance flights. Surface winds are routinely measured by the SFMR (Uhlhorn et al. 2006). The GPS sondes measure profiles of temperature, pressure and humidity from flight level to the surface. For each measurement, 10-m values of these quantities are optimally interpolated to a storm-relative grid aligned with the direction of storm motion. The SFMR wind observations are objectively analyzed (Powell and Houston 1996) and interpolated to the same grid as the GPS thermodynamic data. Finally, the SSTs observed by expendable ocean probes on in-storm flights are interpolated to the same grid as the GPS measurements as in Isidore and Lili (Shay and Uhlhorn 2006). Key items in the estimates for heat and moisture fluxes are temperature and specific humidity differences and the bulk transfer coefficient (i.e., ratio of the enthalpy exchange coefficient to surface drag coefficient (discussed above). As most of the recent studies have indicated a leveling off of the surface drag coefficient at 30 m s-1, Emanuel (2003) suggests that this ratio becomes independent of wind speed and that the ratio should be O(1), since below this value intense hurricanes cannot be simulated in the numerical model. Heat exchange coefficients are typically set equal to Cd, which is conservative based on these recent theories. An additional ocean forcing mechanism is the surface precipitation flux (rain rate). As noted above, freshwater input by rain can alter the ocean response both by direct cooling due to rain that has a lower temperature than the SST, and by stabilizing the OML by decreasing the salinity and reducing the mixing rate (Price et al. 1986; Jacob et al. 2006). Estimates of enthalpy fluxes during Isidore and Lili were sensitive to the storm translation speed. In Hurricane Isidore, the peak enthalpy flux ~1.8 kW m-2 is in the right-rear quadrant of the storm due to the high SSTs (~30oC) due to a negligible decrease from pre-storm conditions, especially over the warm LC where ocean cooling was minimal. Although the maximum momentum flux (7 Pa) is in the right-front quadrant, Isidore's wind stress field was highly symmetric as it moved at only 4 m s-1. Estimated maximum surface fluxes in Lili were about 1.4 kW m-2 due in part to the marked asymmetry associated with the faster storm translation speed (7 m s-1) and smaller SSTs by about 1oC. This result highlights how even modest SST differences can effectively alter the surface heat flux in extreme winds.

142

Figure 1.3.17: Integrated along-track variations in the normalized cross-track direction of the surface heat loss (upper curve) observed during Hurricane Isidore (solid) and Lili (dashed) with the uncertainties based on observed ocean and atmospheric data (from Shay and Uhlhorn 2006). Enthalpy fluxes may also be integrated along the track to obtain the cross-track (radial) distributions of net sea surface heat loss. Using the space-to-time conversion in the along-track direction, estimated surface heat losses for Isidore and Lili are shown in Fig. 1.3.17. The surface heat loss in Isidore (~9 kJ cm-2) is almost a factor of two larger than in Lili (~4.5 kJ cm-2) due to the enhanced flux, slower storm speed, and larger horizontal SST gradients along the western side of the Yucatan Strait. Fluxes are responding to the changes in the upper ocean mixed layer budget and OHC during both storms that are moving over a strong current system. Cione and Uhlhorn (2003) argue that it is only inner- core SSTs that the storm responds to if the OHC was held constant. However, OHC is not remaining constant underneath a storm if SSTs are changing by 1° to 2oC. SST (or near-surface temperatures from profilers) represents the surface boundary condition for OHC estimates from profilers and satellite-based retrieval algorithms, and the SST decrease is not only a function of surface heat flux, but also the stress- and shear-induced mixing (as illustrated in Fig. 1.3.1)

1.3.3.6 Sea Spray Over the last decade, Fairall et al. (1994) and Kepert et al. (1999) have been developing a hierarchy of models of the sea spray production at high winds and the subsequent thermodynamic effects of the evaporation of spray on hurricane boundary layers. The three steps in this process are : i) characterization of the size spectrum of droplets produced by the ocean as a function of the forcing (wind speed, stress, wave breaking, etc); ii) computation of the exchanges of heat and moisture between the droplets and an unperturbed near-surface layer structure; and iii) accounting for the ‘subgrid-scale’ distortion of the standard surface layer temperature and relative humidity structure by the droplets (a process referred to as ‘feedback’). The present source strength parameterization is derived from the Fairall-Banner physical sea spray model that predicts the size spectrum of sea spray produced by the ocean in terms of wind speed, surface stress, and wave properties.

143

Figure 1.3.18: The ratio of momentum (cd) to enthalpy (ck) bulk transfer coefficients when the effects of sea spray are included via different specifications of the droplet source strength (figure courtesy of Chris Fairall). By extending Banner et al. (2000) approach, the The Fairall-Banner spectrum has been parameterized into a simple mass flux representation in terms of surface friction velocity. The unperturbed thermodynamic effects are based on integrals of the ratios of thermodynamic and suspension time constants following Andreas. Finally, a diagnostic feedback parameterization has been developed that characterizes how evaporating droplets of various sizes modify the stratification of the air near the surface, which in turn reduces further droplet evaporation but enhances sensible heat flux carried by the droplets. The present form of the parameterization has two tuning coefficients: one that scales the intensity of the source strength, and the other that affects the partitioning of enthalpy flux between sensible and latent heat. As shown in Fig. 1.3.18, the ratios of momentum of enthalpy transfer coefficients are scaled with wind speed for differing choices of the source strength. Recently, this parameterization was implemented in the GFDL model and a version of WRF model. Preliminary tests with Hurricanes Ivan and Isabel showed sensitivity to the sea spray parameterization, but there are dependencies with the non-droplet (direct) transfer specifications in the models. More testing is needed to examine the sensitivity of these results and determine the role of sea spray on the enthalpy fluxes and hence storm intensification. 1.3.4 Summary Significant progress has been made in understanding the basic oceanic and atmospheric processes that occur during TC passage. The need is to isolate fundamental physical processes involved in the interactions through detailed process studies using experimental, empirical, theoretical, and numerical approaches. As demonstrated from new measurements, these approaches are needed to improve predictions of tropical cyclone intensity and structure. Considerable ocean-atmosphere variability occurs over the storm scales that include fundamental length scales such as the radius of maximum winds and another scale that includes the radius to gale-force winds. Here, the fundamental science questions are how the two fluids are coupled through OML and ABL processes, and what are the fundamental time scales of this interaction? These questions are not easily answered as the interactions must be occurring over various time/space scales. For example, one school of thought is that the only important process with respect to the ocean is under the eyewall where ocean cooling has occurred. While it is at the eyewall where the maximum wind and enthalpy fluxes occur, the broad surface circulation over the warm OML also has non-zero

144

fluxes that are contributing thermal energy to the TC. The deeper the OML (and 26oC isotherm depth), more heat (OHC) is available to the storm through the enthalpy fluxes. Notwithstanding, it is not just the magnitude of the OHC, since the depth of the warm water is important to sustaining surface enthalpy fluxes. Process studies need to begin to look at these multiple scale aspects associated with the atmospheric response to ocean forcing. With regard to the oceanic response to the atmospheric forcing, an important missing ingredient in many studies is the role of the forced and background current fields. In addition to aircraft-based sampling by AXCPs and AXCTDs and new profiling floats such as the EMAPEX and the SOLO (during CBLAST), efforts along the southeastern United States are underway to deploy long-range, HF-radars to map the surface currents to 200 km from the coast as part of an integrated ocean observing system (Shay et al. 2006a). Such measurements would not only be invaluable to map the wind-driven surface currents during high winds, but also in the case of phased arrays, to map the directional wave spectra over the domain. These measurements could then be used to not only examine air-sea interactions, but also assess the relative importance of surface wave-current interactions under strong wind conditions in an Eulerian frame of reference. The variability of the surface drag coefficient has received considerable attention over the last five years including the Coupled Boundary Layer an Air-Sea Transfer (CBLAST) program sponsored by the Office of Naval Research. Several treatments have come to the conclusion that there is a leveling off or a saturation values at about 30 m s-1 +/- 3 m s-1. The ratio of the enthalpy coefficient and the drag coefficient is central to air sea fluxes impacting the TC boundary layer. In this context, the relationship between the coupled processes such as wave breaking and the generation of sea spray and how this is linked to localized air-sea fluxes remains a fertile research area. A key element of this topic is the atmospheric response to the oceanic forcing where there seems to be contrasting viewpoints. One argument is that the air-sea interactions are occurring over surface wave (wind-wave) time and space scales and cause significant intensity changes by more than a category due to very large surface drag coefficients. Yet empirical studies suggest the values to be between 2.5 to 3.4 x 10-3 compared to recent coupled model studies. While, these sub-mesoscale phenomena may affect the enthalpy fluxes, the first-order balances are primarily between the atmospheric and oceanic mixed layers. 1.3.5 Recommendations Based on an Air-Sea Interaction Workshop at NCEP during May 2005 (see http://iwave.rsmas.miami.edu/~nick/AS_HWRF_wksp_rev.pdf), the recommendations are: i.) Given the range of uncertainty in the surface drag (e.g., wave effects), heat fluxes (e.g., sea spray), and initial conditions (e.g., wind field) beyond 30 m s-1 using CBLAST data, assess how these combined uncertainties propagate through the coupled ocean-hurricane model; ii.) Develop an archive of data sets and model outputs and make these archives publicly available for research and operational purposes. Investigate the potential use of these data sets in assimilation, evaluation, and verification of models (e.g., HYCOM) and parameterization schemes; iii.) Create an in-situ tropical cyclone ocean-atmosphere observing program for pre-storm, storm, and post-storm environments. Develop optimal observing strategies and observational mixes for spatial evolution of upper ocean, interface, and atmospheric fields (including secondary circulations such as roll vortices in the hurricane boundary layer); and; iv.) Develop improved ocean model initialization schemes through data assimilation of satellite and in situ measurements, and test mixing parameterizations for a spectrum of ocean, wave and atmospheric conditions including the impact of waves on the surface heat, moisture, and momentum fluxes, and thus on the evolving OML.

145

A common theme in these discussions was that national and international research and forecasting programs need to build on recent field programs (ONR-CBLAST, NSF-sponsored measurements on Hurricane Isidore and Lili) and acquire in situ data over a broad spectrum of atmospheric and oceanographic forcing regimes. These data are needed to test models and examine the parameter space in these air-sea interaction and vertical mixing schemes. Ultimately, future research initiatives must now have strong experimental, empirical, analytical, and numerical modeling components to further our understanding of these fairly complex coupling processes between these two fluids. Acknowledgments: L. K. Shay gratefully acknowledges support from the NSF through grants ATM-01-08218, 04-44525 and NOAA Joint Hurricane Testbed program. We are grateful for the efforts of the pilots, technicians, engineers and scientists at the NOAA Aircraft Operation Center (Dr. Jim McFadden) who make it possible to acquire high quality data during hurricanes. Mr. Bill Teague (NRL-Stennis) and Dr. Rick Lumpkin (NOAA-AOML) kindly shared data and Drs. George Halliwell, S. Daniel Jacob, Tom Sanford and Chris Fairall shared preprint material used herein. Dr. Russell Elsberry made editorial comments that were incorporated by Mrs. Penny Jones at the Naval Postgraduate School. Finally, I appreciate the assistance and guidance of Dr. Hugh Willoughby in preparing this document. References: Banner, M.L., Babanin, A.V., and I.R. Young, 2000: Breaking probability for dominant waves on the sea surface. J. Phys. Oceanogr., 30, 3145-3160. Bao, J.-W., J. M. Wilczak, J. K. Choi, and L. H. Kantha, 2000: Numerical simulations of air-sea interaction under high wind conditions using a coupled model: A study of hurricane development. Mon. Wea. Rev., 128, 2190-2210. Boyer, T.P., and S. Levitus, 1997: Objective analyses and temperature and salinity for the world ocean on a ¼ degree grid. NOAA NESDIS Atlas 11, U.S. Gov. Printing Office, Washington, D.C. Bleck, R., 2002: An oceanic general circulation framed in hybrid isopycnic-Cartesian coordinates. Ocean Modelling, 4, 55-88. Chassignet, E. P., L. Smith, G. R. Halliwell, and R. Bleck, 2003: North Atlantic simulations with the hybrid coordinate ocean model (HYCOM): Impact of the vertical coordinate choice and resolution, reference density, and thermobaricity. J. Phys. Oceanogr., 33, 2504-2526. Chang, S., and R. Anthes, 1978: Numerical simulations of the ocean’s nonlinear baroclinic response to translating hurricanes. J. Phys. Oceanogr., 8, 468-480. Charnock, H., 1955: Wind stress on a water surface. Quart. J. Royal Meteor. Soc., 639-640. Cheney, R., L. Miller, R. Agreen, N. Doyle, and J. Lillibridge, 1994:, TOPEX/Poseidon: The 2-cm solution. J. Geophys. Res., 99, 24,555-24,563. Chérubin, L.M., W. Sturges, and E.P. Chassignet, 2005: Deep flow variability in the vicinity of the Yucatan Straits from a high-resolution numerical simulation, J. Geophys. Res., 110, C04009, doi:10.1029/2004JC002280. Cione, J.J., and E.W. Uhlhorn, 2003: Sea surface temperature variability in hurricanes: Implications with respect to intensity change. Mon. Wea. Rev., 131, 1783-1796.

146

Cooper, M., and K. Haines, 1996: Altimetric assimilation with water conservation properties. J. Geophys. Res., 101, 1059-1077. Cummings, J. A, 2003: Ocean data assimilation, In COAMPS: Version 3 Model Description, NRL Publication NRL/PU/7500-03-448. Naval Research Laboratory, Washington, DC, 20375, 21-28. D’Asaro, E. A., 2003: The ocean boundary layer under hurricane Dennis. J. Phys. Oceanogr., 33, 561-579. DeMaria, M., M. Mainelli, L. K. Shay, J. Knaff, and J. Kaplan, 2005: Further improvements to the statistical hurricane intensity prediction scheme. Wea and Forecasting, 20, 531-543. Donelan, M.A., 1990: Air-sea interaction. The Sea: Ocean Engineering Science, 9, 239-292, 1990. Donelan, M. A., B. K. Haus, N. Reul, W. J. Plant, M. Stiassine, H. Graber, O. Brown, and E. Saltzman, 2004: On the limiting aerodynamic roughness of the ocean in very strong winds. Geophys. Res. Letters., 31L18306,doi:1029/2004GRL019460. Elliot, B. A., 1982: Anticyclonic rings in the Gulf of Mexico, J. Phys. Oceanogr., 12, 1292-1309. Emanuel, K A., 1995: Sensitivity of tropical cyclones to surface exchange and a revised steady-state model incorporating eye dynamics. J. Atmos. Sci. , 52, 3969-3976. Emanuel, K. A., 1999: Thermodynamic control of hurricane intensity. Nature, 401, 665-669. Emanuel, K.A., 2003: A similarity hypothesis for air-sea exchange at extreme wind speeds. J. Atmos. Sci., 60, 1420-1428. Faber, T., L. K. Shay, S. D. Jacob, S. H. Houston, and P. G. Black, 1997: Observed air-sea interactions during hurricane Emily. Preprints, 22nd Conf. on Hurricanes and Tropical Meteorology, Fort Collins, CO, Amer. Meteor. Soc., 433-434. Fairall, C. W., J. D. Kepert, and G. J. Holland, 1994: The effect of sea spray on surface energy transports over the ocean. Global Atmos. Ocean Sys., 2, 121-142. Falkovich, A., I. Ginis, and S. Lord, 2005: Implementation of data assimilation and ocean initialization for the coupled GFDL/URI hurricane prediction system. J. Atmos. and Ocean. Tech., 22, 1918–1932. Fox, D. N., W. J. Teague, C. N. Barron, M. R. Carnes, and C. M. Lee, 2002: The Modular Ocean Data Assimilation System (MODAS). J. Atmos. Oceanogr. Tech., 19, 240-252. Gaspar, P., 1988: Modeling the seasonal cycle of the upper ocean. J. Phys. Oceanogr., 18, 161-180. Gill, A. E., 1984: On the behavior of internal waves in the wakes of storms. J. Phys. Oceanogr., 14, 1129-1151. Goni, G. J., S. Kamholz, S. L. Garzoli, and D. B. Olson, 1996: Dynamics of the Brazil/Malvinas confluence based on inverted echo sounders and altimetry. J. Geophys. Res., 95, 22103-22120. Ginis I., A.P. Khain, B. Thomas, and E. Morozovsky, 2004: Effects of large eddies on the structure of the marine boundary layer under strong wind conditions. J. Atmos. Sci., 61, 3049-3064. Halliwell, G. R., 2004: Evaluation of vertical coordinate and vertical mixing algorithms in the hybrid-coordinate ocean model (HYCOM). Ocean Modelling, 7, 285-322.

147

Halliwell, G., L. K. Shay, E. Uhlhorn, S. D. Jacob, O, O, Smedstad, and C. Lozano, 2006: Initializing ocean models with GODAE ocean nowcast products for tropical cyclone forecasting. J. Atmos. Oceanogr. Tech. (To Be Submitted) Hara, T., and S. E. Belcher, 2004: Wind profile and drag coefficient over mature ocean surface wave spectra. J. Phys. Oceanogr., 34, 2345-2358. Hong, X., S. W. Chang, S. Raman, L. K. Shay, and R. Hodur, 2000: The interaction of hurricane Opal (1995) and a warm core ring in the Gulf of Mexico. Mon. Wea. Rev., 128, 1347-1365. Hock, T. J., and J. L. Franklin, 1999: The NCAR GPS dropwindsonde. Bull. Amer. Meteor. Soc., 80, 407-420. Jacob, D. S., L. K. Shay, A. J. Mariano, and P. G. Black, 2000: The three-dimensional mixed layer heat balance during hurricane Gilbert. J. Phys. Oceanogr., 30, 1407-1429. Jacob, S. D., and L. K. Shay, 2003: The role of oceanic mesoscale features on the tropical cyclone- induced mixed layer response. J. Phys. Oceanogr., 33, 649-676. Jena, B., M. V. Rao and B. K. Sahu, 2006: TRMM-derived sea surface temperature in the wake of a cyclonic storm over the central Bay of Bengal. Inter. J. of Rem. Sens., 27(14), 3065-3072. Kepert, J. D., C. W. Fairall and J.-W. Bao, 1999: Modelling the interaction between the atmospheric boundary layer and evaporating sea spray droplets. Air-Sea Exchange: Physics, Chemistry and Dynamics. G. L. Geernaert, Ed., Kluwer, 369-409. Kitaigorodskii, S.A., 1968: On the calculation of the aerodynamic roughness of the sea surface. Izvestia. Acad. Sci. USSR, Atmos. Oceanic Phys., 4, 498-502. Large, W.G., and S. Pond, 1981: Open ocean momentum flux measurements in moderate to strong winds. J. Phys. Oceanogr., 11, 324-336, 1981. Large, W., S., J. C. McWilliams, and S. C. Doney, 1994: Oceanic vertical mixing: A review and a model with non-local boundary layer parameterization. Rev. Geophys., 32, 363-403. Leaman, K. D., 1976: Observations of vertical polarization and energy flux of near-inertial waves. J. Phys. Oceanogr., 6, 894-908. Leipper, D., and D. Volgenau, 1972: Hurricane heat potential of the Gulf of Mexico. J. Phys. Oceanogr., 2, 218-224. Lin, I.-I., C.-C. Wu, K. A. Emanuel, I.-H. Lee, C.-R. Wu, and I.-F. Pun, 2005: The interaction of supertyphoon Maemi (2003) with a warm ocean eddy. Mon Wea. Rev., 133, 2635-2649. Mainelli-Huber, M., 2000: On the role of the upper ocean in tropical cyclone intensity change. M.S. Thesis, University of Miami, Miami, FL, 73 pp. Mainelli, M., M. DeMaria, L. K. Shay and G. Goni, 2006: Application of oceanic heat content estimation to operational forecasting of recent category 5 hurricanes. Wea. and Forecast. (In Revision) Mariano, A.J., and O.B. Brown, 1992: Efficient objective analysis of heterogeneous and nonstationary fields via parameter matrix. Deep Sea Res., 7, 1255-1271.

148

Marks, F., and L.K. Shay, 1998: Landfalling tropical cyclones: Forecast problems and associated research opportunities: Report of the 5th Prospectus Development Team to the U.S. Weather Research Program, Bull. Amer. Meteor. Soc., 79, 305-323. Mellor, G. L., and T. Yamada, 1982: Development of a turbulence closure model for geophysical fluid problems. Rev. Geophys. and Space Phys., 20, 851-875. Moon I.-J., T. Hara, I. Ginis, S. E. Belcher, and H. Tolman, 2004a: Effect of surface waves on air–sea momentum exchange. Part I: Effect of mature and growing seas. J. Atmos. Sci., 61, 2321–2333. Moon I.-J., I. Ginis, and T. Hara, 2004b: Effect of surface waves on air–sea momentum exchange. II: Behavior of drag coefficient under tropical cyclones. J. Atmos. Sci., 61, 2334–2348. Moon, I., I. Ginis, and T. Hara, 2004c: Effect of surface waves on Charnock coefficient under tropical cyclones. Geophys. Res. Lett., 31, L20302. Nof, D.,2005:, The momentum imbalance paradox revisited, J. Phys. Oceanogr., 35, 1928-1939. Nof, D., and T. Pichevin (2001), The ballooning of outflows, J. Phys. Oceanogr., 31, 3045-3058. Ocampo-Torres, F.J., M.A. Donelan, N. Merzi, and F. Jia, 1994: Laboratory measurements of mass transfer of carbon dioxide and water vapour for smooth and rough flow conditions. Tellus, 46B, 16-32. Ooyama, V., 1969: Numerical simulation of the life cycle of tropical cyclones. J. Atmos. Sci., 26, 3-40. Powell, M.D., P.J. Vickery, and T.A. Reinhold, 2003: Reduced drag coefficient for high wind speeds in tropical cyclones. Nature, 422, 279-283. Price, J.F., 1981: Upper ocean response to a hurricane. J. Phys. Oceanogr., 11, 153-175. Price, J. F., R. A. Weller, and R. Pinkel, 1986: Diurnal cycling: Observations and models of the upper ocean response to diurnal heating, cooling and wind-mixing. J. Geophys. Res.,7, 8411-8427. Price, J. F., T. B. Sanford, and G.Z. Forristall, 1994: Observations and simulations of the forced response to moving hurricanes. J. Phys. Oceanogr., 24, 233-260. Pudov, V., and S. Petrichenko, 2000: Trail of a typhoon in the salinity field of the ocean upper layer. Atmospheric and Ocean Physics (Izvestiya Acad. Sci.), 36, 700-706. Rogers, R., S. Aberson, M. Black, P. Black, J. Cione, P. Dodge, J. Dunion, J. Gamache, J. Kaplan, M. Powell, L. K. Shay, N. Surgi, E. Uhlhorn, 2006: The intensity forecasting experiment (IFEX), a NOAA multiple year field program for improving intensity forecasts. BAMS (In Press). Rosenthal, S. L., 1971: The response of a tropical cyclone model to variations in boundary layer parameters, initial conditions, lateral boundaries and domain size. Mon. Wea. Rev., 99, 767-777. Sanford, T.B., P.G. Black, J. Haustein, J.W. Feeney, G. Z. Forristall, and J.F. Price, 1987: Ocean response to a hurricane. Part I: Observations. J. Phys. Oceanogr., 17, 2065-2083. Sanford, T. B., E. A. D’Asaro, J. B. Girton, J. F. Price, and D. C. Webb, 2006: Water property and momentum changes during the passage of Hurricane Frances. 27th Conf. Hurricanes and Tropical Meteorology, Monterey, CA, 24-28 April 2006, Amer. Meteor. Soc.

149

Scharroo, R., W.H. Smith, and J.L. Lillibridge, 2005: Satellite altimetry and the intensification of Hurricane Katrina. EOS, 86, 366-367. Shay, L. K. 2001: Upper ocean structure: Response to strong forcing events. In: Encyclopedia of Ocean Sciences, Eds. R.A. Weller, S.A. Thorpe, J. Steele, Academic Press International, London, UK, 3100-3114. Shay, L. K., 2002: Air-ocean interface and boundary processes (Chapter 1.4). Topic Chair and Rapporteur’s Report for Tropical Cyclone Structure and Structure Change of the 5th WMO International Workshop on Tropical Cyclones (IWTC-5) 3-12 Dec 2002, Cairns, Australia, World Meteorological Organization Tropical Meteorology Research Series, Ed. Russell L. Elsberry, WMO Report 67, (WMO TD 1136), Geneva, Switzerland. Shay, L.K., and S.W. Chang, 1997: Free surface effects on the near-inertial ocean current response to a hurricane: A revisit. J. Phys. Oceanogr., 27, 23 - 39. Shay, L. K, and S. D. Jacob, 2006: Relationship between oceanic energy fluxes and surface winds during tropical cyclone passage (Chapter 5). Atmosphere-Ocean Interactions II, Advances in Fluid Mechanics. Ed. W. Perrie, WIT Press, Southampton, UK, 115-142.

Shay, L. K., and E.W.Uhlhorn, 2006: Loop Current response to Hurricanes Isidore and Lili. Mon Wea. Rev., (In preparation) Shay, L. K., G. J. Goni, and P. G. Black, 2000: Effects of a warm oceanic feature on hurricane Opal. Mon. Wea. Rev., 128, 1366-1383. Shay, L. K., B. Jaimes, E. Uhlhorn, J. Brewster, M. Mainelli, J. Lillibridge, and P. Black, 2006: Loop Current and warm core ring interactions during hurricanes Katrina and Rita. Geophys. Res. Letters, (Submitted) Shay, L. K., J. Martinez-Pedraja, T. M. Cook, B. K. Haus, and R. H. Weisberg, 2006a: High frequency radar surface current mapping using WERA. J. Atmos. Oceanogr. Tech. (In Press) Sun, D..R. Gautam, G. Cervone, Z. Boyeyi, M. Kaptos, 2006: Comment on Satellite altimetry and the intensification of Hurricane Katrina. EOS, 87(8), 89 Sturges, W., and R. Leben, 2000: Frequency of ring separations from the Loop Current in the Gulf of Mexico: A revised estimate. J. Phys. Oceanogr., 30, 1814-1819. Teague, W. J., M. J. Carron, and P. J. Hogan, 1990: A comparison between the generalized digital environmental model and Levitus climatologies. J. Geophys. Res., 95, 99-116. Teague, W. J., D. A. Mitchell, E. Jarosz, and P. J. Hogan, 2005: Low-frequency current variability observed at the shelf break in the northeastern Gulf of Mexico: May - October, 2004. Cont. Shelf Res. (Submitted). Tolman, H. L., 2002: User manual and system documentation of WAVEWATCH-III version 2.22. NOAA/NWS/NCEP/OMB Tech. Note 222, 133 pp. Uhlhorn, E. W., and L.K. Shay, 2004: Analysis of upper-ocean thermodynamic observations forced by Hurricane Lili. 26th Conference on Hurricanes and Tropical Meteorology, 3-7 May, Miami Beach, FL, Amer. Meteor. Soc., 619-620.

150

Uhlhorn, E. W., and P. G. Black, 2003: Verification of remotely sensed sea surface winds in hurricanes. J. Atmos. Ocean. Tech, 20, 99-116. Uhlhorn, E. W., P. G. Black, J. L. Franklin, M. Goodberlet, J. Carswell and A. S. Goldstein, 2006:Hurricane surface wind measurements from an operational stepped frequency microwave radiometer. Mon. Wea. Rev. (Submitted) Walker, N., R. R. Leben, and S. Balasubramanian 2005: Hurricane forced upwelling and chlorophyll a enhancement within cold core cyclones in the Gulf of Mexico. Geophys. Res. Letter, 32, L18610, doi: 10.1029/2005GL023716, 1-5. Wang, Y., J. D. Kepert, and G. J. Holland, 2001: The effect of sea spray evaporation on tropical cyclone boundary layer structure and intensity. Mon. Wea. Rev., 129, 2481-2500. Wright, C. W., E. J. Walsh, D. Vandemark, W. B. Krabill, A. W. Garcia, S. Houston, M. Powell, P. Black, and F. D. Marks, 2001: Hurricane directional wave spectrum spatial variations in the open ocean. J. Phys. Oceanogr., 31, 2472-2488. Wu, C.-C., C.-Y. Lee, and I-I Lin, 2006: The effect of the ocean eddy on tropical cyclone intensity. J. Atmos. Sci. (Submitted)

151

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES

Topic 1.4 : Operational Techniques in Defining TC Structure Rapporteur: Mark A. Lander University of Guam UOG Station Mangilao, Guam (USA) 96923 Email: mlander@uog.edu Fax: 1-671-734-8890 Working Group: A. Zhao, C.-S. Liou, K. Cheung, R. Edson, and J. Franklin Abstract:

Tropical cyclone warnings to the public typically include forecasts of the weather elements that contribute to hazards that may threaten lives or cause substantial damage to infrastructure, homes, crops and other physical assets. Traditionally, the TC warning issued to the public contains information on the expected wind speeds, rainfall amounts, and sea inundation. There is often a broad-brush statement on the level of damage to be expected. TC structures that are important to the operational forecast and warning efforts include: (1) surface wind distribution; (2) eye characteristics; (3) vertical structure of the wind and temperature; (4) Satellite-observed cloud patterns; (5) critical boundary layer profiles (e.g., changes as TC moves over colder water) (6) sub-kilometer scale wind patterns in the TC core that might have a substantial impact on air-sea

exchanges. Even at the most sophisticated of operational centers, one might consider the techniques for determining TC structure (e.g., wind distribution and intensity estimation) fairly primitive. The techniques for estimating TC intensity from satellite imagery have changed little in 30 years. There is, however, a vast array of new sensors and applications that should greatly improve the ability of forecasters to diagnose the TC structures that are critical to the making of accurate forecasts and improved public warnings. This report highlights some of the key technologies that promise great advances in operational ability to diagnose TC structure. The TC community must convey to policy makers the high priority of these technologies. There are many sensors that help the forecaster to diagnose TC structure. The conventional visible and infrared satellite pictures are the “bread and butter” of TC structure diagnosis, and will continue to be the anchor sensors for determining TC structure. 1.4.1: Overview The primary characteristic of a tropical cyclone that is used to assess and anticipate the threat to human life and property is the intensity, which is normally given as the maximum wind speed in the TC core. The Saffir-Simpson Hurricane Damage Potential Scale is one of the few resources available to relate a forecast wind speed in a TC to expected levels of damage. The minimum sea level pressure in the TC center is also used as a measure of intensity. Lowered surface pressure in the TC core,

152

however, is not a primary hazard, but does contribute to coastal inundation through the inverse barometric rise of sea level. Other structural characteristics of a TC that have operational importance in the preparation of forecasts and warnings include: (1) the surface wind distribution; (2) the eye characteristics (e.g., concentric or annular); (3) the vertical structure of the wind and temperature; and, (4) the spatial extent and organization of deep convection; Small-scale wind transients such as TC-associated tornadoes are also important structural features of TCs. Other small-scale characteristics of a TC may have important implications for local wind conditions; for example, the boundary layer temperature profile and changes to this profile such as those occurring when a TC moves over colder water. Some esoteric characteristics of the TC core such as the sub-kilometer scale wind depiction provided by high-resolution Doppler radar may prove to be beneficial for numerical forecasts of high-impact TC structures, such as the intensity. 1.4.2. The TC surface wind distribution For many years, conventional visible (VIS) and infrared (IR) satellite imagery has provided the input data for the diagnosis of TC intensity and wind distribution. Dvorak’s techniques (developed in the early 1970s) became operational with the introduction of his techniques for estimating TC intensity from visible satellite imagery (Dvorak 1975), and the techniques for estimating TC intensity from enhanced infrared satellite imagery (Dvorak 1984). Aircraft reconnaissance of TCs is routinely available only in the Atlantic basin. Many of the operationally important structural characteristics of TCs can be obtained from the aircraft platform; otherwise, meteorological satellites, synoptic observations and ground-based radar (where available) must be used to determine TC structure. The diagnosis of TC surface wind distribution has not seen substantial improvement until recent advances occurred in remote-sensing technologies, including active microwave sensors on satellites and aircraft. Historical efforts to diagnose TC wind distribution have primarily rested on determination of the size of the TC from conventional satellite imagery. The outer wind distribution is then obtained using an analytical TC wind profile (e.g., Holland 1980) anchored by the estimated size and intensity. There are typically five bins of TC size (very small, small, average, large, and giant) cf Brand (1972) and Merrill (1982). At the Joint Typhoon Warning Center (JTWC) Guam (in Hawaii after 1998), the values in Table 1 were used – in the absence of other evidence to the contrary – to assign wind radii to TCs. This table was used at the JTWC at least through the mid 1990s (Edson, personal communication). Of course, any other reliable data depicting the actual wind radii of a given TC trumped the values obtained in Table 1 – especially wind vectors obtained by scatterometer. Other considerations used to fine-tune the diagnosis and forecasts of TC wind distribution included: the TC movement; the synoptic environment; the diameter of the eye (especially as revealed by passive microwave imagery); land effects; and extratropical transition. Powell et al. (1998) developed a technique (the “H*wind” software) to derive the surface wind speeds in a TC. H*Wind is an analysis package that, like any other analysis scheme, takes observations at specific locations (like along aircraft legs) and deduces a complete surface wind-field depiction for the entire domain of the TC. Forecasters at the NHC weren't completely satisfied with the accuracy and consistency of the technique. Some contentious issues included the appropriateness of presenting to the public a depiction of the surface winds in the entire TC when the wind field was sampled only in limited areas; and the propriety of adjusting flight-level winds downward to deduce a surface wind distribution over the entire cyclone. For reasons such as these, it was not implemented operationally (Franklin, personal communication). At the NHC, one might consider the techniques for determining the wind distribution fairly primitive. Traditionally the flight-level winds are printed out in hard copy format, and dividers are used on these

153

plots to estimate the operational wind radii (64, 50, 34 kt radii in the 4 quadrants). When this is done, a standard set of surface wind adjustment factors derived from dropsonde data are used. Very recently, the NHC gained the technical capacity to locally ingest the flight-level winds into N-AWIPS, thus one can display the recon winds along with all the other observations (ships, buoys, METARs, QuikSCAT, etc.) on one screen with the satellite imagery. There are distance tools in N-AWIPS, so it is likely that this is how operational wind radii will be obtained until further advances are forthcoming. N-AWIPS can also adjust the flight-level wind speeds using the dropsonde adjustment factors. Just recently (early September 2006) SFMR ingest was implemented. The NHC is moving rapidly toward a single display platform for all of these data. The use of airborne radar to depict structure operationally is in its infancy. There is a Joint Hurricane Testbed (JHT) project ongoing in the U.S. to develop displays of vertical and horizontal reflectivity and Doppler velocity data that could be used by the forecaster to assess some important TC structural characteristics (e.g., banding, concentric eyewall, wind distribution, etc.) None of this has quite made it into operations yet -- maybe within a year or two. 1.4.3 Eye Characteristics One of the most important structural features of a TC is the character of the eye. Operational techniques for determining TC intensity, TC wind distribution, and other metrics (such as the radius of maximum wind) depend on accurate knowledge of the eye structure. Dvorak’s techniques for determining TC intensity from satellite imagery are heavily dependent on the structure of the eye. Advances in remote sensing -- in particular, microwave imagers -- now enable forecasters to see details of eye structure through dense cirrus. The implications of the properties of the TC eye as seen on microwave imagery are only now beginning to be understood. Two features of eye structure have received recent operational attention: (1) Eyewall replacement cycles (Willoughby et al. 1982 and Willoughby 1990), and (2)annular hurricanes (Knaff et al. 2003). (a) Concentric eyewalls "Concentric eyewall cycles" (or "eyewall replacement cycles") naturally occur in intense tropical cyclones, i.e. major hurricanes (winds > 50 m/s, 100 kt, 115 mph) or Catories 3, 4, and 5 on the Saffir-Simpson scale. During the replacement phase, the tropical cyclone typically weakens. Eventually the outer eyewall replaces the inner one completely and the storm can be the same intensity as it was previously or, in some cases, even stronger. A TC may undergo more than one concentric eyewall cycle. For example, Hurricane Allen (1980) went through repeated eyewall replacement cycles -- going from Categrory 5 to Category 3 status several times. Microwave imagery greatly enhances the ability of the forecaster to ascertain concentric eyewalls and ongoing replacement cycles. (b) Annular Hurricanes An annular hurricane is a tropical cyclone that features a large, symmetric eye (Fig. 1) surrounded by a thick ring of intense convection. This type of TC is not prone to the fluctuations in intensity associated with eyewall replacement cycles, unlike typical intense TCs. Forecasters have difficulty predicting the behavior of annular hurricanes; they are a relatively recently recognized phenomenon. Annular hurricanes are axisymmetric — symmetric along every radial axis, i.e. very circular in appearance. They lack the spiralform rainbands that are characteristic of typical TCs. After reaching peak intensity, they weaken much more slowly than non-annular storms of similar intensity. However, most annular hurricanes have annular characteristics for only a portion of their lifetimes. Annular hurricanes maintain their intensities longer than usual after their peaks. Statistics show that forecasters significantly overestimate the lessening of wind velocities in annular hurricanes. In terms of the Dvorak technique, annular hurricanes weaken very slowly after their peak (on average, less than 0.5 T after one day from their peak intensities). Research into the characteristics and formation of annular hurricanes is still in its infancy. First classified and categorized in 2002, little is known about how they form, or why some are able to maintain their intensity in hostile conditions. What meteorologists do know is that a normal hurricane, after undergoing an eyewall replacement cycle, fails to re-establish the

154

standard hurricane appearance. The new eyewall thickens, and rainbands dissipate, and the hurricane takes on an annular structure. As compared to the formation of normal hurricanes, this happens under weaker wind shear and, surprisingly, cooler sea surface temperatures. 1.4.4 Vertical Structure of the wind and temperature One of the most vexing operational decisions is whether or not to issue TC advisories on cyclones that may not possess all of the characteristics that are normally thought to define a TC. The so-called subtropical cyclone is a prime example. Monsoon depressions in the western Pacific and in the Indian Ocean also fall into this problematic category of cyclones that do not possess the features common to “normal” TCs. Beginning in 2002, the U.S. National Hurricane Center began to number and name subtropical cyclones. In addition, the numbering and naming of the subtropical cyclones follow the natural sequence of the numbering and naming system for the tropical cyclones. Hebert and Poteat (1975) provided a description of the characteristics of the subtropical cyclone that still applies today. The subtropical cyclone typically has only a weak lower tropospheric warm-core structure resulting from the lack of sustained convection near the cyclone center. A large radius of maximum winds is associated with these cyclones. To obtain the satellite classification of a subtropical cyclone is a simple process of looking at a series of satellite pictures of subtropical cyclones of various intensities and choosing the one that best applies to the case in question. The NHC strategy has worked well, and all of the named subtropical cyclones since 2002 have made the transition to tropical (though this need not be the case). Recent work by Hart (2002) gives a quantitative way to track a cyclone through a 3-D phase space based on the thermal wind in the core of the cyclone in the lower-troposphere, the thermal wind in the core of the cyclone in the upper troposphere, and the horizontal temperature gradient in a section drawn across the core of the cyclone. The previously sharp boundary distinguishing tropical cyclones from extratropical cyclones in the first half of the 20th century has been substantially weakened. Techniques for recovering the vertical temperature profile and the sea level pressure in tropical and subtropical cyclones from satellite-observed microwave emission has helped the forecaster to assess the nature of a particular cyclone. The Advanced Microwave Sounding Unit (AMSU) provides operationally useful information about TC structure. The Microwave Sounding Unit (MSU) began service in 1978 on TIROS-N and continued on the NOAA 6 through 14 satellites. AMSU flies on the NOAA KLM and N satellites as well as NASA's Aqua. There is strong relationship between the brightness temperature anomalies as measured by the AMSU-A instrument and TC intensity (http://amsu.ssec.wisc.edu/explanation.html). The CIMSS AMSU algorithm uses this relationship to estimate TC MSLP. In general during the early stages of TC development the associated warm core is located near channel 7 and that channel is used to produce an estimate. As the TC intensifies the warm core moves higher in the atmosphere closer to the mean location of channel 8. Experience indicates that once the TC reaches hurricane intensity channel 8 tends to be the better indicator of storm strength and the algorithm uses that channel. 1.4.5 Summary of new instrumentation for observing TC structure Microwave sensors offer a chance to greatly improve the diagnosis of TC structure. These sensors are both active (e.g., QuikSCAT, the TRMM precipitation radar, and the stepped-frequency microwave radiometer used on hurricane reconnaissance aircraft) and passive (e.g., the TRMM microwave imager and the microwave imagers onboard many polar-orbiting satellites). These technologies provide data that should greatly improve the ability of forecasters to diagnose the TC structures that are critical to the making of accurate forecasts and improved public warnings.

155

(a) QuikSCAT The QuikSCAT satellite is a polar-orbiting, sun-synchronous satellite with an equator-ascending time of approximately 0600 local (+/- 30 minutes). Wind vector solutions and up to four ambiguity solutions are stored in 25 km X 25 km wind vector cells over an 1800 km wide swath. Besides including the position of the center of each cell, each cell also contains various flagging indicators that include possible contamination with rain, land, and ice. Combining the QuikSCAT data with that of other recent satellite-based remote sensing data can provide a clear and very accurate analysis of TC position and outer wind radii from the genesis phase all the way through to the extratropical transition phase of development, and finally provide a ‘minimum’ (at least) maximum intensity for those systems just at or below hurricane intensity. In a special case, the high resolution NRCS image can reveal the structure of the TC during an eyewall replacement cycle. Forecasters were initially reluctant to use the Quikscat ocean surface wind products (e.g., http://manati.orbit.nesdis.noaa.gov/quikscat/ and http://www.nrlmry.navy.mil/sat-bin/scatt_winds.cgi) to diagnose TC surface wind structure because of the perception that some of the problems in the wind retrievals (e.g., effects of rain and wind directional ambiguities) rendered the product unfit, or too difficult to interpret. However, relentless efforts by Edson (e.g., Edson 2002, Edson and Chang 2003, and Edson and Lander 2003) and others (Liou and Jin 2006) revealed that the QuikSCAT wind retrievals and the raw data stream from the satellite (i.e., the normalized radar cross section – NRCS) could readily be interpreted to yield invaluable information about TC structure. (b) The Stepped-Frequency Microwave Radiometer (SFMR) Measurement of the hurricane surface wind field, and in particular the estimation of wind maxima, has long been a requirement of the Tropical Prediction Center/National Hurricane Center (TPC/NHC). The NOAA/Hurricane Research Division's (HRD) Stepped-Frequency Microwave Radiometer (SFMR), built by Prosensing Inc., is the prototype for a new generation of airborne remote sensing instruments designed for operational surface wind estimation in hurricanes. The first experimental SFMR surface wind measurements were made in Hurricane Allen in 1980, the first real-time retrieval of winds on board the aircraft in Hurricane Earl in 1985, and the first operational transmission of winds to TPC/NHC in Hurricane Dennis in 1999. Uhlhorn and Black (2003) determined that surface winds measured by the SFMR are comparable to the Global Positioning Systems (GPS) dropwindsonde measurements that are the current standard. GPS dropwindsondes are instrument packages designed to measure wind speed, temperature and humidity as they drop from the aircraft to the surface. The benefit of the SFMR is that winds are continuously measured during research flights, allowing for more complete mapping of hurricane surface wind structure. The active 2005 hurricane season in the Atlantic offered several opportunities to test the SFMR algorithm under extreme wind conditions. Details of The Extreme Wind SFMR Algorithm Adjustment, 2005 are found at http://www.aoml.noaa.gov/hrd/project2005/katrinasfmr.html. NOAA WP-3D flights into Hurricanes Katrina and Rita provided the first opportunity to confirm the ability of the pod-mounted production version of the SFMR to measure extreme surface wind speeds, i.e. surface wind speeds in excess of 120 kt. It appears that over the entire range of hurricane force winds (i.e. > 64 kt), the emissivity/windspeed relationship may in fact be linear, as opposed to the quadratic behavior previously employed. One can see that the effect of the new fit is to remove the observed small (< 5kt) high bias in the mid range of wind speeds, i.e. 50-100 kt. At the same time the high-wind linear fit also corrects the low bias at extreme winds, i.e. > 115 kt, that had been interpreted incorrectly as a high wind 'roll-off'. Further, it can be concluded that the slight high wind bias in the mid-range was due to the erroneous choice of a quadratic algorithm extending to high winds. (c) The TRMM precipitation radar The active radar aboard the TRMM satellite, the TRMM precipitation radar (TRMM-PR), has proven useful in the diagnosis of TC structure. Kodama and Yamada (2005) examined the eye detectability and configuration of 138 TC cases in the western North Pacific during 1998-2002 using satellite

156

infrared (IR) imageries and the TRMM-PR. It was found that TC eyes were detectable in PR data in 89% of cases and in IR data in only 37% of cases. The cases when the eye was detected both in IR and PR data are mostly the TCs with stronger intensity. In most cases, eye diameter was smaller in IR observations than in PR observations because an upper cloud shield extending from the eyewall partially covered the eye. Cecil et al. (2002) examined 261 TRMM satellite overpasses of 45 hurricanes during 1997-1998. The study focused on the distribution of radar reflectivity values, passive microwave ice scattering magnitudes, and total lightning among the eyewall, inner rainband and outer rainband. It was found that for nearly half of the hurricane’s raining area, measurable radar echo (> 17 dB) does not extend more than 2 km above freezing level. The median echo top extends to 8 km in the convective rainbands and up to 9 km in the eyewall. Most of the rain areas are stratiform with only 9% of the eyewall region meets the “convective certain” criteria. In addition, lightning flash densities are four times greater in the eyewall region and outer rainband region than in the inner rainband region, which is partially attributed to the stratiform nature of the inner rainband region. (d) Other sensors This report highlights QuikSCAT and the SFMR as technologies that promise great advances in operational ability to diagnose TC structure. The TC community must convey to policy makers the high priority of these two technologies. There are many other sensors that help the forecaster to diagnose TC structure: Conventional Visible and Infrared satellite imagery, passive microwave imagery, and ground-based radar. The conventional visible and infrared satellite pictures are the “bread and butter” of TC structure diagnosis, and will continue to be the anchor sensors for determining TC structure. TC reconnaissance aircraft will likely remain a tool for only the most affluent, but smaller unmanned aerial vehicles (UAVs) may provide a less expensive alternative for in situ monitoring of TCs. One such UAV, the Aerosonde (http://www.aerosonde.com/) is scheduled in 2006 for single or multiple experiments coordinated with NOAA and AFRES aircraft missions. Since 2003, the Dropsonde Observations for Typhoon Surveillance near the TAiwan Regions (DOTSTAR; Wu et al. 2005) program has conducted 19 typhoon surveillance missions with the Astra aircraft in North-western Pacific. For some target typhoons, the impact of dropwindsonde data on mesoscale weather prediction was investigated. Experiments were made to test the impact of the subset of the dropwindsonde data, Taiwan terrain, vortex bogusing, and data assimilation schemes. REFERENCES Brand, S., 1972: Very large and very small typhoons of the western North Pacific Ocean. J. Meteor. Soc. Japan, 50, 332-341. Cecil, D. J., E. J. Zipser, and S. W. Nesbitt, 2002: Reflectivity, ice scattering, and lightning characteristics of hurricane eyewalls and rainbands. Part I: Quantitative description. Mon. Wea. Rev., 130, 769−784. Dvorak, V., 1975: Tropical cyclone intensity analysis and forecasting from satellite imagery. Mon. Wea. Rev., 103, 420-430. Dvorak, V., 1984: Tropical cyclone intensity analysis using satellite data. NOAA Tech. Report NESDIS 11. Available from NOAA/NESDIS, 5200 Auth Rd., Washington DC, 20233, 47pp. Edson, R.T., 2002. Tutorial on QUIKSCAT. Special Focus Topic 1.b. Proceedings of the Fifth WMO International Workshop on Tropical Cyclones (IWTC-V), Cairns, Queensland, Australia. 3-12 December 2002. WMO/TD.

157

Edson, R.T. and P.S. Chang, 2003 : Normalized radar cross-section patterns from QuikSCAT—A new analysis tool over the tropical ocean. Proc of the 12th Conf on Satellite Meteor. and Oceanography, Long Beach, CA. Edson, R.T. and M.A. Lander, 2003: A method for integrated satellite reconnaissance fix accuracy. Proceedings of the 12th Conference on Satellite Meteorology and Oceanography, Long Beach, CA. Hart, R.E., 2003: A cyclone phase space derived from thermal wind and thermal asymmetry. Mon. Wea. Rev., 131, 585–616. Hebert, P.H., and K.O. Poteat, 1975: A satellite classification technique for subtropical cyclones. NOAA Tech. Memo. NWS SR-83, 25 pp. Holland G.J., 1980: An analytical model of the wind and pressure profiles in hurricanes. Mon. Wea. Rev., 108, 1212-1218. Huntley, J.E., 1980: Tropical Cyclone Wind Radius Program. Annual Tropical Cyclone Report of the Joint Typhoon Warning Center, Guam. p 121. NTIS Acquisition Number AD A094668. Knaff,J.A., J.P. Kossin, and M. DeMaria 2003: Annular Hurricanes. Weather and Forecasting, 18, 204-223. Kodama, Y.-M., and T. Yamada, 2005: Detectability and configuration of tropical cyclone eyes over the western North Pacific in TRMM PR and IR observations. Mon. Wea. Rev., 133, 2213-2226. Liou C.-S. and Y. Jin, 2004: A Method of Applying QuikSCAT Data for Tropical Cyclone Initialization, Proceedings of the 26th Conference on Hurricanes and Tropical Meteorology, Miami, FL. Merrill, R.T., 1984: A comparison of large and small tropical cyclones. Mon. Wea. Rev., 112, 1408-1418. Powell, M. D., S. H. Houston, L. R. Amat, N. Morisseau-Leroy, 1998: The HRD real-time hurricane wind analysis system. J. Wind Engineer. Ind. Aerody., 77&78, 53-64. Uhlhorn, E.W., and P.G. Black, 2003: Verification of Remotely Sensed Sea Surface Winds in Hurricanes. Journal of Atmospheric and Oceanic Technology, 20, 99–116. Willoughby, H.E., J.A. Clos, and M.G. Shoreibah, 1982: "Concentric eye walls, secondary wind maxima, and the evolution of the hurricane vortex" J. Atmos. Sci., 39, pp.395-411. Willoughby, H.E. 1990: "Temporal changes of the primary circulation in tropical cyclones" J. Atmos. Sci., 47, pp.242-264 Wu, C.-C., P.-H. Lin, S. Aberson, T._C. Yeh, W.-P. Huang, K.-H. Chou, J.-S. Hong, G.-C. Lu, C.-T. Fong, K.-C. Hsu, I-I Lin, P.-L. Lin, and C.-H. Liu, 2005: Dropwindsonde Observations for Typhoon Surveillance near the Taiwan Region (DOTSTAR): An Overview. Bull. Amer. Meteor. Soc., 86, 787-790.

158

Figure 1. Hurricane Isabel of 2003 showing annular hurricane structure. Notice the large eye (partially filled by eyewall mesovortices) and the relatively few spiral bands around the outside of the storm.

159

Table 1. Reconstructed wind-radii table used at the JTWC (Sampson and Edson, personal communication)

PREDICTED WIND RADII FOR SPECIFIED WIND INTENSITY (HUNTLEY, 1980, BASED ON TC VORTEX PROFILES BY G. HOLLAND)

RADIUS (NM) WIND (KT) VERY SMALL

RMW = 10 SMALL

RMW = 15 AVERAGE RMW = 20

LARGE RMW = 25

VERY LARGE RMW = 30

WIND RADII FOR XX KT WIND RADII FOR XX KT WIND RADII FOR XX KT WIND RADII FOR XX KT WIND RADII FOR XX KT

100 64 50 34 100 64 50 34 100 64 50 34 100 64 50 34 100 64 50 34 INTENSITY

0 0 0 30 0 0 0 45 0 0 0 60 0 0 0 70 0 0 0 75 35kts 0 0 0 35 0 0 0 55 0 0 0 70 0 0 0 85 0 0 0 95 40kts 0 0 0 40 0 0 0 65 0 0 0 80 0 0 0 100 0 0 0 120 45kts 0 0 15 45 0 0 20 75 0 0 30 90 0 0 35 115 0 0 45 140 50kts 0 0 20 50 0 0 30 85 0 0 40 100 0 0 50 130 0 0 60 160 55kts 0 0 25 60 0 0 40 95 0 0 55 130 0 0 65 155 0 0 80 190 60kts 0 20 30 70 0 35 50 115 0 45 70 150 0 55 85 185 0 65 100 220 65kts 0 20 35 80 0 35 55 125 0 50 80 170 0 60 95 200 0 70 115 245 70kts 0 25 45 90 0 40 65 135 0 55 90 180 0 65 110 220 0 80 130 270 75kts 0 30 50 100 0 40 70 140 0 60 100 195 0 70 120 240 0 85 145 290 80kts 0 30 55 105 0 45 80 155 0 60 105 210 0 75 130 255 0 90 155 310 85kts 0 30 60 110 0 45 85 165 0 65 110 215 0 75 140 270 0 95 165 325 90kts 0 35 65 120 0 50 90 170 0 65 120 225 0 80 150 280 0 100 175 340 95kts

15 35 65 120 15 50 95 175 20 70 125 235 25 85 155 290 35 105 185 355 100kts 15 35 70 125 20 50 100 185 25 70 130 245 30 90 160 300 40 110 195 370 105kt 15 35 70 125 20 50 105 190 30 70 135 250 40 90 165 310 50 110 200 375 110kts 20 40 70 130 25 55 105 190 35 75 140 255 45 95 170 315 55 120 210 390 115kts 20 40 75 135 30 60 110 195 40 80 145 260 50 100 175 320 65 130 215 395 120kts 20 40 75 135 30 65 115 200 45 90 150 265 55 105 180 325 70 135 220 400 125kts 25 45 75 135 35 70 115 205 45 90 150 270 60 110 185 330 75 140 225 410 130kts 25 45 80 140 35 70 120 210 50 95 155 280 65 115 190 335 80 145 230 415 135kts 25 50 80 145 40 75 120 210 55 100 160 285 70 120 195 340 80 145 235 420 140kts 30 50 80 145 40 75 120 215 55 100 160 290 70 125 200 350 85 150 240 425 145kts 30 50 85 150 45 80 125 215 60 105 165 295 75 130 205 355 85 150 245 430 150kts 30 50 85 150 45 80 125 220 65 110 170 300 80 135 210 360 95 160 255 440 160kts 35 55 90 155 50 85 130 225 70 115 175 305 85 140 215 365 100 165 260 450 170kts

160

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES Topic 1.5 : Operational guidance and skill in forecasting structure change Rapporteur: John Knaff Research Scientist NOAA Cooperative Institute for Research in the Atmosphere Colorado State University Fort Collins, CO 80523-1375 USA Email: Knaff@cira.colostate.edu Telephone: 970 491-8881 Working Group: C. Guard, J. Kossin, T. Marchok, B. Sampson, T. Smith, N. Surgi Abstract: Currently operational tropical cyclone structure change forecasts consist of the forecast of maximum intensity in terms of maximum surface winds and the radial extent of winds exceeding various wind thresholds, commonly 34-kt, 50-kt, and 64-kt. A survey of operational forecasters suggests that the process of making intensity and wind radii forecasts has changed little since IWTC-5. During the same time, verification shows that intensity guidance has been steadily improving, albeit slowly, and is driving operational forecast improvements. This result was determined by using the historical databases available in the Automated Tropical Cyclone Forecast System and conducting a long-term verification of operational intensity forecasts and intensity guidance methods. Two metrics are used to verifying intensity change forecasting including the traditional measure of mean absolute errors (MAE) and the percent reduction in variance of the observed intensity change. Findings show that MAEs have very small decreasing trends and the percent reduction of variance has small increasing trends. Guidance for wind radii forecasting is currently not skillful and the best wind radii guidance is produced by statistical methods based on climatology and/or persistence. After examining the verification results, there is a clear need for continued tropical cyclone structure change guidance improvement and a few topics related to ongoing and future research to improve operational forecasting of TC structure are discussed. 1.5.1 Introduction In the operational setting there are several aspects of TC structure that are analyzed and forecast. The maximum intensity has been long analyzed and forecast by operational centers. The intensity is often defined in terms of the maximum sustained wind (MSW) at 10-m over a time averaging period (1, 3, 10- minute), which varies by operational center or by the minimum sea level pressure (MSLP). To complement these TC metrics of intensity, several quantities that describe the structure of the TC vortex near the surface are also analyzed at each advisory period. These vary from operational center to operational center and include: the radius of maximum winds (RMW), eye diameter, the radius of outer closed (closed and circular) isobar (OCI), and the maximum extent of wind speed in quadrants (e.g., 34-kt, 50-kt and 64-kt wind radii). With the exception of OCI, these quantities describe the near surface wind field. A summary of the operational determination of structural aspects of the TC are provided in Section 1.4. The MSW is directly related to the potential impact of the TC. This quantity has been long estimated and forecast by operational TC forecasting centers. The relatively long operational history of this quantity stems from the ability to estimate this quantity from satellite imagery and aircraft

161

measured/estimated MSLP. In regions without routine aircraft reconnaissance, the primary technique to operationally estimate MSW was developed by Dvorak (1975, 1984), but more recently other techniques (Velden et al. 1998, Cocks et al, 1999; Bruske and Velden 2003; Demuth et al. 2004; 2006, Olander and Velden 2006) have aided these estimates. In operational centers that have access to aircraft reconnaissance, MSW is estimated using flight-level reduction, dropwindsonde observations, surface wind observations, MSLP observations (via wind-pressure relationships), and more recently, observations from an operational Stepped Frequency Microwave Radiometer (SFMR; Uhlhorn et al 2006). Despite the accuracy/precision and sometimes uncertainty associated with the various MSW estimate methods (Brown and Franklin 2002; 2004; Velden et al. 1998; Olander and Velden 2006), these MSW estimates provide a long history of observed structural variability of tropical cyclones. Because of this long operational history, this quantity is the primary metric of tropical cyclone structure and is forecast and verified at all operational TC centers. While MSW is related to the potential destruction of a given TC near the region of strongest winds, it is the size of the wind field that is, in many instances, best related to the total impact of the TC. The extent or arrival time of strong winds (e.g., gale force winds) is very important for making pre-storm preparations by coastal residents, government agencies and other concerned parties. The relative size of the wind field is also important in the determination of other coastal impacts such as storm surge, and wave setup. For these reasons, some operational centers forecast the radial extent of various wind thresholds. The most common observed metric is the extent of gale force (34-kt) winds. Mostly because of the difficulty in observing and therefore verifying the wind field associated with tropical cyclones, only a few operational centers provide forecasts of the wind field. As track and intensity forecasts have become better, there has been more emphasis on TC wind structure. Since the last IWTC, wind radii estimates have become part of the annual best track at Regional Specialized Meteorological Centre (RSMC), Tokyo, RSMC, Miami, RSMC, Honolulu and the Joint Typhoon Warning Center. Such information has lead to the recent development of simple models to predict the structure of the TC vortex. These simple models enable the verification of numerical weather prediction (NWP) wind field guidance and operational forecasts, which will be discussed here. The topic of this report is the operational capabilities available to forecast the aspects of TC structure. In the next section a review of the guidance available and its use in preparing forecasts of TC structure (i.e., intensity and wind radii) will be presented. The following section will present a verification of intensity and wind radii forecasts made by both operational centers and by their guidance methods. A discussion of future needs, issues and directions along with a summary will follow. 1.5.2 TC Structure Forecast Guidance

a) Operational intensity change guidance

All operational TC forecasting centers issue MSW or intensity forecasts, but the guidance that is used in these forecasts varies considerably. There are several types of intensity guidance, including:

1. 24-h forecasts based on Dvorak (1984), 2. Purely statistical models developed from historical data, 3. Statistical-dynamical models which make use of environmental information from NWP models,

climatology, persistence and satellite-derived data to make statistical forecasts 4. forecasts from NWP models

In addition to these guidance techniques operational centers make use of other predictive and diagnostic indices to aid in the intensity forecasting process. An often utilized guidance technique is the 24-h forecast described in Dvorak (1984) whereby the

162

forecast intensity is a forward extrapolation of the past 24-h change in T-number (not to exceed 1.5 T-number per day). This extrapolation is then modified by a set of rules related to the TC’s cloud pattern and environment. These Dvorak forecasts are still used as the primary method for predicting 24-h TC intensity change at many operational centers. Another common guidance technique is the use of purely statistical models. An example of one of these models is the SHIFOR model (Jarvinen and Neumann 1979) used at the RSMC, Miami, which has produces 3-day intensity forecasts from current location and intensity,12-h trends in intensity, and motion. Some of these simple models use purely climatological information from analogs (Sampson et al. 1990), while others employ the combined aspects of climatology and persistence (Chu et al. 1994; Jarvinen and Neumann 1979; Knaff et al. 2003). Such models can be used to make operational forecasts, however their primary role is as a skill reference during verification. In verification, these types of models provide a means to normalize forecasts that are more difficult than climatology or a combination of climatology and persistence. Often forecasts are considered skillful if they outperform climatology and persistence based forecasts. One potential drawback of using such models as benchmarks for intensity verification is that they do not take into account the effects of landfall. Thus models that take landfall into consideration when making intensity forecasts can gain skill through this effect (James Franklin, personal communication). Caution therefore should be used in interpreting the level of skill determined from this type of intensity verification for cases affected by land. In the last 10 years or so, statistical-dynamical approaches have been developed to predict intensity change. These models make use of environmental information from NWP models’ forecast fields, and SSTs along the forecast track along with information derived from climatology and persistence. Traditionally these models have been developed using a “perfect prog” assumption where NWP analyses and best track positions are used in the model development. The first and most advanced version of these is the Statistical Hurricane Intensity Prediction Scheme (SHIPS), which now makes forecasts every 6 hours through 5 days along the official forecast track and makes use of environmental information from US National Centers for Environmental Prediction (NCEP) Global Forecast System (GFS) model along with information from weekly SSTs, infrared satellite imagery, and ocean heat content estimated from satellite-based altimeters (DeMaria et al. 2005a). Forecasts are created for the N. Atlantic, and combined Eastern and Central Pacific regions by separate versions of the SHIPS model created specifically for each basin. Another set of operational models include the Statistical Typhoon Intensity Prediction Scheme (STIPS) which makes 12-hourly forecasts through five days along the official forecast track (Knaff et al 2005). The STIPS model is currently run at the Joint Typhoon Warning Center (JTWC) and the US Naval Research Laboratory for all JTWC’s areas of responsibility and a consensus/ensemble (consemble) of STIPS forecasts along a variety of forecast tracks and using a number of different NWP forecast was initiated in operations during 2005 with successful results (Sampson et. al. 2006). There are several other statistical-dynamical guidance models that have been developed that provide experimental guidance to operations, including the passive microwave version of SHIPS and STIPS (Cecil et al. 2006), and a Neural Network based model (Baik et al 2003). Many of these types of models adjust landfalling forecasts by the employment of one of a number of inland decay models (Kaplan and Demaria 1995; 2001; DeMaria et al. 2006). These statistical-dynamical models produce forecasts that are available either during the forecast cycle or as early guidance. Intensity guidance is also provided from a variety of global, regional and specialized NWP models. The global NWP models US Navy Operational Global Analysis and Prediction System (NOGAPS; Hogan and Rosmond 1991; Goerss and Jeffries 1994), Japanese global spectral model (JGSM;Kuma 1996), the NCEP Global Forecast System (GFS; Lord 1993), United Kingdom Meteorological Office (UKMO) global model (Cullen 1993; Heming et al. 1995) are operationally available for intensity prediction. Because of the relatively limited spatial resolutions of the global models, it is common that intensity estimates from these models are created by adding the intensity change forecast by these models to the observed initial intensity as part of the interpolation process discussed below. To complement the global models there are a number of regional and specialized NWP models which

163

have finer spatial resolutions. These include the US Geophysical Fluid Dynamics Laboratory hurricane model (GFDL; Kurihara et al. 1993; 1995; 1998), GFDN (GFDL with a NOGAPS initialization; Rennick 1999), the Japanese typhoon model (JTYM; Kuma 1996), the US Navy’s Coupled Ocean/Atmosphere Mesoscale Prediction System (COAMPS®; Hodur 1997), Australia’s TC-Limited Area Prediction System (TC-LAPS; Davidson and Weber 2000), the fifth-generation Pennsylvania State University/National Center for Atmospheric Research Mesoscale Model (MM5; Grell et al. 1995) run operationally by the Air Force Weather Agency (AFWA). The NWP models take a number of different approaches to initialize the hurricane vortex. For example, the Geophysical Fluid Dynamics Laboratory (GFDL) hurricane model, one of the most advanced regional models, uses output from the GFS for initial and lateral boundary conditions; however it removes the coarse-resolution vortex found in the global model analysis fields. The GFDL system utilizes near-surface wind radii observations provided by RSMC, Miami to generate a 3-dimensional bogus vortex which is then inserted into the initial fields. Other regional models use similar wind radii - based initialization (MM5, GFDN) while others are initialized using the pressure field (e.g., TC-LAPS and JTYM). Currently, the operational centers provide the information used to create the model initialization. Improved methods to initialize tropical cyclone vortices in NWP models are an area of active research and development. Most models also assume that the SST fields remain constant throughout the forecast integration, except for the GFDL hurricane model. In 2001, the GFDL model was coupled with a 3-dimensional ocean model (Bender and Ginis, 2000) thus including the effects of turbulent ocean mixing and upwelling. The primary benefits of the ocean coupling are to simulate the cold ocean wake left behind the storm and to inhibit the over-development of tropical cyclones that move slowly over very warm waters. The coupling of atmospheric and oceanic physics has been shown to improve intensity forecasts. It is important to note that the intensity guidance from the NWP models, both regional (e.g. GFDL, JTYM etc.) and global (e.g., NOGAPS, UKM, etc.), arrives to the RSMCs and other operational centers too late in order to be used as guidance for the current synoptic cycle’s forecast package. For example, the 12 UTC run of the GFDL model does not finish until a few hours after the 12 UTC forecast package has been issued by RSMC, Miami. In order to maximize the utility of the NWP model forecasts, interpolated versions of forecasts from these models are created by calculating the difference between the 6-h forecast intensity and the observed intensity at that hour and then adding that difference to the forecast intensity at all subsequent forecast hours. These interpolated results are often referred to as early guidance. In addition to traditional guidance that provides a deterministic estimate of TC intensity, there are a few predictive and diagnostic indices that provide probabilistic information related directly or indirectly to intensity change. An operational example of such an index is the rapid intensification index (Kaplan and DeMaria 2003), which provides the probability of rapid intensification (increase of 30 kt or greater in 24 h) in the next 24 h. Other indices that relate to structural or environmental changes that may lead to intensity changes are the Secondary Eyewall Formation Index (SEFI; Kossin et. al.,cited 2006), the Annular Hurricane Index (AHI; Cram et al. 2006) and storm relative shear tendency (Gallina and Velden 2002). The SEFI uses information from the environment and from passive microwave imagery along with a Bayes Classifier algorithm to predict the probability of secondary eyewall formation. Secondary eyewall formation is shown to cause short-term fluctuations in TC intensity. The AHI determines the probability of a given TC being annular. Annular hurricanes have been found to have a stable structure and maintain their current intensity longer than non-annular hurricanes (Knaff et al. 2003; Cram et al. 2006). The shear tendency is calculated from satellite feature drift winds and has been routinely emailed to operational offices. The SEFI and AHI are to be tested during the later half of 2006 in the Atlantic and East Pacific.

164

b) Shortcoming of Operational Intensity Guidance Intensity guidance has a number of shortcomings. Purely statistical models do not take environmental factors or landfall into consideration and as a result have very conservative forecasts. Statistical-dynamical models also have a conservative nature built into them. Like the purely statistical models, these models are based primarily on multiple linear regressions and will predict the mean of the sample. As a result, these models cannot predict rapid changes in intensity. In addition, current statistical-dynamical models employ time averaging and as a result, large environmental changes have little effect at long forecast leads. Global NWP does not have the spatial resolution necessary to capture the intensification process, whereas regional/specialized NWP models often suffer from poor vortex initialization that affects their short-term forecasts. Because of operational time constraints, all of the current operational models rely on physical parameterizations for the smaller scale processes (convection, turbulence, boundary-layer, ice microphysics etc.). The continued and necessary use of parameterizations, including convective parameterization, handicaps current NWP. With the exception of the GFDL hurricane model, the effects of turbulent ocean mixing and upwelling cannot affect intensity change, leading in some cases to over-development of tropical cyclones moving slowly over warm waters. Finally, all guidance suffers from errors associated with timing of tropical cyclone intensification as the process of intensification is still an area of active research and model development (see Booth et al 2006; Blackerby 2005, and Lambert 2005 for details).

c) How Intensity Change Guidance is Used Operational intensity forecasting remains, as it always has been, a somewhat subjective exercise. Unlike track forecasting, intensity forecasting involves a complex interaction between many spatial and temporal scales ranging from convective to synoptic. These interactions and the incomplete understanding of the structural change process limit the ability and utility of operational intensity change forecasts. As a result, operational intensity forecasts are a blend of the guidance and the judgment of the operational forecaster. In this respect, the best operational guidance serves as a baseline on which the subjective forecast is based. Thus it follows that if the best guidance or the baseline improves, so should the operational forecasts. This blend of guidance and human judgment varies from operational center to center as well as from situation to situation. At many operational centers the number of intensity guidance tools is rather limited and even when available the guidance has limited utility. The short-term (first 24-hours) forecasts of intensity are most often based on recent intensity trends that are modified by ongoing or anticipated changes in the environment and the storm, which come from a blend of satellite interpretation, and numerical model forecast fields. This procedure is either identical or very similar to the method proposed by Dvorak (1984) for making 24-h intensity forecasts and has changed little in the last ten years, while the availability and quality of satellite imagery and model analyses/forecasts of the environment has greatly improved. Changes in atmospheric moisture, vertical wind shear, sea surface temperature, and outflow conditions are often examined in a storm relative manner to help predict the timing and magnitude of intensity change. Forecasters also examine the current structure of the tropical cyclone for convective structure (e.g. for evidence of vertical wind shear, concentric eyewalls, eyewall formation/dissipation, etc.), and trends in convective symmetry and strength. Simple statistical models as well as statistical-dynamical models aid in this process by serving as a baseline or first guess. As the forecast lead increases to 36 through 72 hours, there is a greater reliance on intensity trends in NWP, modified by subjective analysis of the forecast environment changes and trends in the statistical–dynamic guidance. At the farthest lead times, 5-day at some operational forecast centers, forecasts are often relaxed toward climatology for that situation.

d) Operational wind radii guidance, its use and shortcomings. Some operational centers provide forecasts of wind radii. These forecasts are a subjective blend of guidance and common sense forecasting. All wind radii forecasts start with a current assessment of

165

the tropical cyclone’s wind structure. An important ingredient to this assessment is satellite imagery and products, particularly in those regions without aircraft reconnaissance data. Since there are rarely enough ship, bouy or synoptic observations to estimate the wind structure, scatterometry (QuickScat, NScat etc.), which allows the estimation of the gale force winds, has become a vital tool to the operational forecaster. The reliance on scatterometry is important to mention as few new instruments of this type are planned in the future. Other in-house techniques that rely on infrared and passive microwave imagery have been and continue to be used at various operational centers. These techniques simply relate features in the imagery to storm size and with the addition of an intensity estimate provide a wind profile estimate. An example of such a method is the Holland & Martin Technique (Martin 1990). Some forecasters also utilize analyses from NWP models, but because of poor representation of the storm vortex of many models their utility is limited. Finally, in some instances the forecaster will rely on climatological vortex structure from in-house programs, tabular forms, nominal values of wind radii, and from analysis from high resolution NWP. These initial conditions along with anticipated changes of storm intensity and the environment form the basis for most wind radii forecasts. Further details of methods used to evaluate tropical cyclone structure are discussed in Section 1.4.

To address some of the subjective nature associated with these initial wind radii estimates and the distinct possibility of no future scatterometry, new techniques have been developed and are working their way into operations. Initial wind radii estimates are now also being provided by a number of recently developed techniques that use microwave sounders (Demuth et al. 2004; 2006) and infrared imagery (Kossin et al. 2006; Mueller et al 2006). Current wind radii guidance is very limited, most of which are very simple methods. These include:

1. Climatologies in tabular or equation form as a function of the intensity forecast 2. Purely statistical models that make forecasts based on climatology and persistence 3. NWP guidance 4. Climatological means

The utility of each of these methods is also limited. Climatology has the longest history. There are several programs, tables and equations used in forecast offices that provide a first estimate of wind radii. One example is the Huntley model used at the JTWC for several years based, though documentation on the tables used in the program is elusive (Cocks and Gray 2002). Some experienced forecasters use their own climatologies developed from personal experience while others rely on parametric model estimates of various wind radii as a function of intensity. Of the potential guidance methods, only NWP can account for complex interactions with the environment. In saying this, it should also be noted that NWP models often poorly represent the wind field due to model resolution and poor initialization. And even when the NWP model can represent the vortex, an accurate intensity forecast is often needed to assign wind radii at future times. In the last couple of years, statistical models have been developed in some of the basins that predict wind radii based on climatology and persistence (Knaff et al 2006; McAdie 2004). In operations these models make forecasts of wind radii based on the initial wind radii conditions, the forecast track, and the forecast intensity. At this time, few forecasters use such guidance in preparing their forecasts as it is a rather new tool, but these are being used as benchmarks to evaluate and verify wind radii forecasts from other models. Based on a quick survey of operational forecasters, the use of climatologies stratified by size is a useful and often employed method of making forecasts. Size is often assessed from the initial conditions as described above (also Section 1.4). Often forecasts of wind radii are derived by modifying a blend of initial conditions and climatology according to the concurrent intensity forecast. Asymmetries in these forecast wind radii are routinely added to account for translation speed, synoptic conditions (e.g., interaction with westerly flow), gradient changes associated with landfall, and vortex structure (i.e. tilting associated with vertical wind shear). Higher resolution numerical models are also sometimes

166

utilized for the qualitative assessment of size of wind radii and their extent and asymmetries. For instance, a forecaster in the Brisbane Tropical Cyclone Centre will sometimes make use of the ECMWF wind fields assigning gales where the 850 hPa wind is 45 kt and similarly the GFDL model may be used by a forecaster in Miami. One thing is certain, wind radii forecasting is a very subjective activity which varies from forecast center to forecast center and from forecaster to forecaster.

e) Guidance for outer closed isobar To be complete in this report on forecasting and verifying tropical cyclone structure, there should be some mention of the operational use of outer closed isobar. The quantity of outer closed isobar has been long used as a tropical cyclone size parameter. In fact, quite a few studies concerned with understanding and forecasting tropical cyclone size changes (e.g. Merrill 1984; Weatherford and Gray 1988; Cock and Gray 2002) have used this parameter. The outer closed isobar is often used along with a surface pressure estimate to initialize some regional and global models (e.g., Japan Typhoon Model and TC-LAPS). Recently it has been found that the outer closed circular isobar, which is routinely provided by Australian forecast centres, produces an improved vortex initialization in TC-LAPS (Harry Weber, personal communication). At this time no operational center is forecasting the change of this quantity, but there is a long history of the use of this size parameter in operational centers from which future techniques could be developed. 1.5.3 TC Structure Verification

The previous section discussed the guidance available for making tropical cyclone structure forecasts as well as a summary of how this guidance is utilized to make operational forecasts of intensity and wind radii. This section will try to address three issues:

1. How good are operational forecasts of TC intensity and structure? 2. How good are the guidance methods available to the operational centers? 3. How have the intensity guidance and forecasts changed over time?

The intensity and wind radii forecasts and best track information used in these verifications come from the ATCF (Sampson and Schrader 2000) databases archived at the RSMC, Miami and the Joint Typhoon Warning Center (JTWC). This will allow the comparison of intensity forecasting in four primary basins; the North Atlantic, the East Pacific, the northwest Pacific and the Southern Hemisphere. These are then compared with other published verifications from the RSMCs. There are two methods used to verify intensity forecasts. The first is the traditional measure of Mean Absolute Error (MAE) of the wind speed forecasts. The second is to determine the percent reduction in variance (PRIV) by the intensity forecasts. This quantity is calculated by subtracting the ratio of the sum of the square intensity errors to the variance of the intensity change from the value of 1 as shown in Eq. 1, where o is the observed intensity change and p is the predicted intensity change and the overbar represents a mean value.

−

−−=

∑

∑N

n

N

nn

oo

poPRIV

1

2

1

2

)(

)(0.1100 (1)

The variance of intensity change can be thought of as the square errors associated with a climatological forecast (i.e., the mean intensity change for the sample). Thus if the forecast errors are greater than those produced by climatology the reduction of variance can be negative (i.e., worst than

167

climatology). Three aspects of the wind radii forecast will be examined. They are the probability of detection (or Hit Rate), the False Alarm Rate (Mason and Graham 1999) and the mean absolute errors (in units of n mi). The Hit rate is defined as the ratio of the number of times specific wind radii (e.g., R34) are forecast to exist to the number of times those wind radii are observed to exist. The false alarm rate is the ratio of the number of times wind radii are forecast to the number of cases when wind radii were not observed.

a) Operational intensity change verification.

Traditionally intensity error verification has been expressed in terms of mean absolute error (MAE) or root mean square error (RMSE). The post-season reanalyized or “best track” intensity estimates are used for this verification. Historical (1986-2005) MAEs associated with intensity forecasts in the North Atlantic, East Pacific (1990-2005) as forecast by RSMC, Miami (i.e., NOAA/TPC), West Pacific and Southern Hemisphere (1991-2005) as forecast by the Joint Typhoon Warning Center (JTWC) are presented in Fig. 1.5.1. These intensity errors are consistent with those reported by other operational forecast centers. Table 1.5.1 shows the 2004-2005 official forecast errors produced by RSMC, La Reunion and Table 1.5.2 shows the official forecast errors reported by RSMC, Tokyo. These historical values of MAE show little improvement has been made in the last 20 years. The MAE associated with intensity forecasts improved only slightly in three of the basins, and actually increased over time in the Southern Hemisphere. Table 1.5.3 shows the observed trends of the intensity errors in units of knots (1 kt = .514 ms-1) per decade by forecast hour and basin. These trends are marginally significant (p>.80) using a Student’s t Test, except in the Southern Hemisphere. The largest downward trends are observed at the longer lead times with little improvement at 24 hours.

N. Atlantic

5

10

15

20

25

30

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Year

Mea

n A

bsol

ute

Erro

r (k

t)

244872

E. Pacific

5

10

15

20

25

30

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Year

Mea

n A

bsol

ute

Erro

r (k

t)

244872

168

Southern Hemisphere

5

10

15

20

25

30

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Year

Mea

n A

bsol

ute

Erro

r (k

t)2448

W. Pacific

5

10

15

20

25

30

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Year

Mea

n A

bsol

ute

Erro

r (k

t)

244872

Figure 1.5.1: Mean absolute errors associated with annual tropical cyclone forecasts of intensity in the North Atlantic, East Pacific as forecast by RSMC, Miami, and Southern Hemisphere and West Pacific tropical cyclone basins as forecast by JTWC. Errors resulting from 24, 48 and 72- hour forecasts are shown by black, red and blue lines. Units are in knots.

1.5.1

2004-2005 seasonal intensity mean absolute errors, root mean square errors and bias in units of kt accumulated at RSMC, La Reunion, which makes forecasts in portions of the South Indian Ocean (Philippe Caroff, personal communication). Range 0h 12h 24h 48h 72h Average error (kt) 3 6 9 14 16 RMSE (kt) 4 6 9 12 13 Bias (kt) -1 -1 -2 -4 -2 Skill against persistence

6% 31% 43% 50%

Sample (number of forecasts verified)

310

303

291

255

213

169

1.5.2

RMSE in ms-1 and kt in parentheses associated with annual intensity forecasts at RSMC, Tokyo, which makes forecasts in the western North Pacific basin (JMA, cited 2006).

24-h 48-h 72-h 2004 5.1 (9.9) 7.1 (13.8) 8.1 (15.8) 2003 4.9 (9.5) 6.5 (12.7) 7.6 (14.8) 2002 5.0 (9.7) 7.0 (13.6) N/A 2001 5.2 (10.1) 6.9 (13.4) N/A 2000 5.9 (11.5) N/A N/A

1.5.3

Trends of the mean absolute intensity forecast in terms of MAE per decade in units of knots for the North Atlantic (ATL), East Pacific (EPAC), Southern Hemisphere (SHEM) and West Pacific (WPAC) at forecast times of 24, 48 and 72 hours. Results based upon forecasts produced by RSMC, Miami (ATL, EPAC) and JTWC (WPAC, SHEM).

24-h 48-h 72-h ATL (1986-2005) -0.8 -1.0 -1.4 EPAC (1990-2005) 0.0 -0.8 -1.9 SHEM (1991-2005) 2.0 2.8 N/A WPAC (1986-2005) -0.2 -0.6 -1.1

The annual percent reduction of the intensity change variance is also examined for these annual forecasts. This analysis shows much greater interannual variability exists in this verification statistic (Fig. 1.5.2). Despite the greater variability, a slow but steady increase in the percent reduction in variance is seen in all basins over the times of record, even in the Southern Hemisphere where the trends in MAEs of intensity forecasts were shown to be increasing. These upward trends in % per decade are shown in Table 1.5.4 and are greatest again at the longer forecast lead times. Trends in the Atlantic are highly significant (P > 0.99) and marginally significant (P > 0.80) in the East Pacific and West Pacific using a Student’s t Test.

170

24-h Variance Explained

-60%

-40%

-20%

0%

20%

40%

60%

80%

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Year

Perc

ent R

educ

tion

in V

aria

nce

WPACATLEPACSHEM

48-h Variance Explained

-60%

-40%

-20%

0%

20%

40%

60%

80%

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Year

Perc

ent R

educ

tion

in V

aria

nce

WPACATLEPACSHEM

72-h Variance Explained

-60%

-40%

-20%

0%

20%

40%

60%

80%

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Year

Perc

ent R

educ

tion

in V

aria

nce

WPACATLEPAC

Figure 1.5.2: Time series of the percent reduction of the variance of tropical cyclone intensity change associated with the JTWC forecasts in the West Pacific (WPAC) and Southern Hemisphere (SHEM) and RSMC, Miami forecasts in the North Atlantic (ATL) and East Pacific (EPAC) for the 24-, 48- and 72-hour forecasts. Note that 72-h forecasts are not currently issued in the Southern Hemisphere.

1.5.4

171

Long-term trends in the percent reduction of variance associated with 24-,48- and 72-hour forecasts in four tropical cyclone basins. Results are given for the North Atlantic (ATL), East Pacific (EPAC), Southern Hemisphere (SHEM) and West Pacific (WPAC) in terms of % increase per decade. Results are based upon forecasts produced by RSMC, Miami (ATL, EPAC) and JTWC (WPAC, SHEM).

24-h 48-h 72-h ATL (1986-2005) 14.7 22.3 41.0 EPAC (1990-2005) 1.5 1.4 3.3 SHEM (1991-2005) 2.0 2.8 N/A WPAC (1986-2005) 3.2 3.0 3.6

The large trends in the Atlantic basin are primarily due to the rather poor forecast performance during the late 1980’s and early 1990’s when the only guidance was from SHIFOR, a purely statistical model. The first versions of the SHIPS model and of the GFDL hurricane model were available to Atlantic forecasters in 1991 and 1992, respectively. Other basins do not show these dramatic changes associated with more guidance. In the East Pacific, the period 1995-1996 saw the number of guidance tools increase as the GFDL hurricane model and the SHIPS model became available in that basin. Similarly in the West Pacific, the GFDN became available in 1996, a 5-day statistical model in 2002, the STIPS model in 2003 and the STIPS consemble in 2005. The Southern Hemisphere has had relatively little change in guidance with only the addition of the GFDN in 2000-2001, the TC-LAPS for the Australian regions in 2004 and a 5-day CLIPER model in 2004. While the number of intensity guidance tools has increased, especially with the addition of intensity forecasts from the global models in the last several years, the quality of the guidance is more important. To determine if the guidance is influencing the forecast the 15-year period 1991-2005 is used to examine the homogeneous verification of the official forecasts and the available guidance. For this verification, the guidance had to be available 60% of the verification period to be considered “available”. Figure 1.5.3 shows the percent reduction of variance for the official (i.e., JTWC or RSMC, Miami) 48-hour forecasts along with the percent reduction of variance of the best guidance tool available to the forecasters at that lead time. Note results are similar for 24-hour and 72-hour forecasts (not shown). These time series show great variability between basins. The East Pacific forecasts, for instance, show relatively large reductions in variance since 1991, whereas the other basins show much larger interannual variability in this statistic, which is particularly apparent in the Atlantic. Also shown is that the percent reduction in variance from the guidance has an upward trend in all basins, especially in the Atlantic and West Pacific.

172

Atlantic

-30%-20%-10%

0%10%20%30%40%50%60%70%80%

1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Year

% R

educ

tion

in V

aria

nce

RMTC, MiamiGuidance

East Pacific

-30%-20%-10%

0%10%20%30%40%50%60%70%80%

1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Year

% R

educ

tion

in V

aria

nce

RMTC, MiamiGuidance

Southern Hemisphere

-30%-20%-10%

0%10%20%30%40%50%60%70%80%

1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Year

% R

educ

tion

in V

aria

nce

JTWCGuidance

West Pacific

-30%-20%-10%

0%10%20%30%40%50%60%70%80%

1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Year

% R

educ

tion

in V

aria

nce

JTWCGuidance

Figure 1.5.3: Homogeneous time series of annual percent reduction in variance associated with 48-hour official forecasts and the best guidance available for that year. Results are shown for the Atlantic, East Pacific, Southern Hemisphere and West Pacific.

173

In tabulating the results shown here, two guidance types have accounted for the increase in forecast ability in the last several years. The GFDL and GFDN models as well as the statistical-dynamical models SHIPS and STIPS have accounted for most of the improvement in guidance in these basins. Despite the efforts to utilize intensity change information from the global models, these models are unable to predict intensity change better than a persistence forecast during most years and are therefore not useful as intensity guidance. In the last couple of years, ensemble and consensus methods to make intensity forecasts have been tested or transitioned to operations. These methods include the STIPS consemble (Sampson et al. 2006) and the Florida State Superensemble (FSSE; Mackey et al., 2005), both of which have been shown to have increased skill above that of existing guidance. An interesting result was presented by Sampson et al (cited 2006) and was also reported in Franklin et al (2006) in that the average results of existing and skillful intensity models (the SHIPS model with inland decay and interpolated GFDL model) outperformed all intensity guidance including the FSSE during 2005, and further improvements can be made by adding the information from the interpolated official forecast. Such results support the idea of using a consensus of skillful intensity forecasting methods in operations. As the guidance has improved, so have the forecasts, though this is not all that evident in the Southern Hemisphere, where guidance has only improved very recently. Thus it appears that improvements in guidance (regional NWP and statistical-dynamical models) since ~2000 have lead to small but steady improvements in operational intensity forecasts.

b) Operational wind radii verification.

It has been just recently that operational centers (RSMC, Tokyo, RSMC, Miami, and JTWC) have been conducting a postseason reanalysis of wind radii. Such reanalysis makes the verification of operational wind radii forecasts and forecast guidance possible. However, the short history of these forecasts and the recent improvement in best tracks does not allow a comprehensive verification of tropical cyclone wind radii like was possible with intensity. Instead of a historical verification, this section will concentrate on the verification of gale force (34-kt) winds of a single well observed year and basin - 2005 in the Atlantic. The verification of wind radii presents new issues. In addition to the MAE associated with the individual forecast methods, there is also the underlying issue of the detection of wind radii, which is a function of intensity. For instance if the 48-h forecast predicts an intensity of 60 kt, 64-kt winds will not be detected. To study this sensitivity to intensity prediction a statistical-parametric model (Knaff et al. 2006) is utilized where one set of forecasts used the official intensity forecast (DRCL) and another set of forecasts used the best track intensities (DRCC). Figure 1.5.4 shows the probability of detection (or Hit Rate) and False Alarm Rate (Mason and Graham 1999) and the mean absolute errors in n mi (1 n mi = 1.85 km) of gale force wind radii in all four quadrants for the official forecast (OFCL) and the two sets of statistical forecasts. Note that if the forecast or the best track has zero for the wind radii, the forecast is verified if the best track intensity exceeds the threshold of the wind radii. A perfect intensity forecast results in a 3 to 11 % increase in the probability of detection and a 15 to 70% decrease in the false alarm rate. According to this analysis, the official forecast is also skillful thru 72 hours, though its lower MAE may partially be due to a larger false alarm rate. As a result, any improvement in intensity forecasting will likely lead to greater detection and better forecasting of wind radii. Note that the mean absolute intensity errors for this sample are 7.4, 11.3, 13,7, 16.0, 20.8, 20.9, and 21.5 kts, with biases of -.7, -1.6, -3.4, -5.4, -7.9, -11.0, -and 11.5 kts at 12, 24, 36, 48, 72, 96,and 120 hours. Using the same verification method used above, the verification the wind radii guidance is now examined. There are four models that are evaluated. They are the GFDL hurricane model (GFDT), the NCEP GFS (AVNI) global model, and two models based on climatology and persistence DRCL (Knaff et al. 2006), MRCL (McAdie 2004). The NWP guidance has been interpolated. Figure 1.5.5 shows the verification statistics associated with these models. It is clear that the NWP-based guidance is inferior to the statistical models. In the case of the AVNI the problem is clearly one of

174

detection suggesting that the intensity estimate is too poor to produce wind radii forecasts directly. GFDT still has problems with detection, but since the GFDL model produces skillful intensity forecasts (Franklin, cited 2006), its detection problem is likely due to other issues related to model resolution or the representation of the vortex in the model. Both statistical models perform well, but there seems to be a trade off with MAE and false alarm rate for these models as was the case with the OFCL forecasts. As the false alarm rate increases, the model MAEs decrease. In summary, this analysis suggests that NWP is unable to better predict wind radii associated with tropical cyclones than simple statistical models. This agrees with the assessment of Knabb (cited 2006) and answers questions posed at IWTC-V. The causes are likely threefold and are a combination of unskillful intensity forecasts, poor vortex initialization and insufficient resolution to capture the correct vortex structure. A possible solution may be to correct the model output statistically to produce improved forecasts.

175

0.5

0.6

0.7

0.8

0.9

1

12-h 24-h 36-h 48-h 72-h 96-h 120-h

Forecast Hour

Hit

Rat

e

OFCLDRCLDRCC

0

0.2

0.4

0.6

0.8

1

12-h 24-h 36-h 48-h 72-h 96-h 120-h

Forecast Hour

Fals

e A

larm

Rat

e

OFCLDRCLDRCC

0102030405060

12-h 24-h 36-h 48-h 72-h 96-h 120-h

Forecast Hour

MA

E (n

mi)

OFCLDRCLDRCC

Figure 1.5.4: Hit rate, False Alarm Rate and MAE [n mi] associated with a homogeneous sample of various forecasts of 34-kt wind radii. Results are shown for the official RSMC, Miami forecast (OFCL), a statistical-parametric model using forecast intensities (DRCL), and that same model using observed intensities (DRCC).

176

0.2

0.4

0.6

0.8

1

12-h 24-h 36-h 48-h 72-h

Forecast Hour

Hit

Rat

e OFCLAVNIGFDTMRCLDRCL

0

0.2

0.4

0.6

0.8

1

12-h 24-h 36-h 48-h 72-h

Forecast Hour

Fals

e A

larm

Rat

e

OFCLAVNIGFDTMRCLDRCL

102540557085

100

12-h 24-h 36-h 48-h 72-h

Forecast Hour

MA

E (n

mi)

OFCLAVNIGFDTMRCLDRCL

Figure 1.5.5: Hit rate, False Alarm Rate and MAE [n mi] associated with a homogeneous sample of various forecasts of 34-kt wind radii. Results are shown for the official RSMC, Miami forecast (OFCL), the NCEP GFS (AVNI), the GFDL hurricane model (GFDT), a statistical-parametric CLIPER model (DRCL) and a purely statistical CLIPER model (MRCL).

177

1.5.4 Summary, recommendations, and directions for the future.

a) Summary

The first part of this paper reviews how tropical cyclone structure, both intensity and wind radii, are forecast in an operational setting and what guidance is available. After surveying operational forecasters and centers, findings suggest that the process of making operational tropical cyclone structure forecasts is still subjective and the degree of subjectivity depends on the quality of the guidance. The best guidance is often used as a baseline for intensity forecasts, which is modified by the forecaster, based on his assessment of the future environmental conditions, knowledge of guidance weaknesses, and her/his experience. The shortest term forecasts are based on observations of recent trends and an assessment of the environment. As the forecast lead time increases beyond 24-h, guidance become more influential in the intensity forecasting process, but only through ~72 hours, where errors in tracks and environmental forecast errors make the guidance less useful. The longest range forecasts are often a blend of the guidance through 72 or 96 hours and a relaxation toward climatology. Wind radii forecasting is even more subjective than intensity forecasting. The process of making wind radii forecasts varies from forecast center to forecast center and from forecaster to forecaster and is dependent on the initial wind radii estimates and the future intensity forecasts. Operational forecast centers use tabular climatologies with respect to intensity, climatological averages, parametric models and statistical models to provide a baseline forecasts. The baseline forecast is then modified to include the effects of storm translation, and synoptic interactions (increasing the asymmetries in the wind field). The verification of intensity forecasts shows that operational forecast mean absolute errors (skill) are decreasing (increasing) very slowly and that the increases are largest at longer leads (see Tables 1.5.3 and 1.5.4). It also appears that the improvements in intensity guidance have improved more rapidly over the last 15 years and are now driving current and future intensity forecast improvements. The improvements in operational guidance, which are resulting in better operational forecasts, come as primarily the result of advancement in statistical-dynamical models and of regional/specialized/mesoscale hurricane models. It is clear that intensity forecasting is more advanced than forecasting wind radii, which is still in its infancy. There is little or no guidance to aid the forecaster in making such forecasts better. The best guidance is in the form of statistical models based on climatology and persistence or climatology alone. All wind radii prediction models are affected by suboptimal intensity forecasts as shown in Figure 1.5.4 where the probability of detection is increase by about 10 percent by just having the correct intensity forecast. Numerical radii guidance suffers from three shortcomings: 1) coarse resolution and inability to correctly represent the vortex structure;2) poor vortex initialization; and 3) unskillful intensity forecasting. This poor performance is highlighted in Figure 1.5.5, which shows the hit rate, false alarm rate, and MAE associated with Atlantic wind radii guidance models during 2005. The verification results suggest tropical cyclone structure is rather poorly forecast. Clearly there is a great opportunity for our field to develop better tropical cyclone structure forecasts whether it be by using existing model output to create new statistical models, by developing new statistical techniques, or by refining numerical weather prediction model capabilities. Clearly the latter will likely be the ultimate solution. In the mean time, other techniques should be also pursued.

b) Future and ongoing research and development

Future and ongoing research and development falls in two categories. The first is to use existing

178

technology and data more efficiently to better diagnose and predict tropical cyclone structure. The second concerns itself with the development of new technology that will lead to improvements in the diagnosis and prediction of tropical cyclone structure. In the first category, there are several items that should be considered for future work, but only a few are discussed. The first is the continued development of new statistical and probabilistic techniques that make use of existing data and technology. Examples include using model output from global and regional NWP models to statistically predict changes in intensity and structure. This is clearly needed as most current NWP models do not have skill in predicting wind radii or intensity. Another example is to develop methods that are designed specifically to predict the short-term (less than 24-h) intensity changes, which by design would address the issue of rapid intensification. Such techniques could leverage the advances in satellite data, the improved environmental analyses, and advanced statistical techniques. Current examples include the Secondary Eyewall Formation Index (Kossin et al, cited 2006), the Annular Hurricane index (Cram et al. 2006) and the Rapid intensity index (Kaplan and DeMaria 2003). Finally, in light of the slow improvement in intensity forecasting and the difficulties in predicting wind radii, there should be a continued emphasis on techniques that convey uncertainty associated with the forecast. A good example of this type of strategy is the Monte Carlo tropical cyclone wind probability model developed for the Atlantic, East/Central and northwest Pacific (Gross 2004; DeMaria et al 2005b) which provides probabilities of gale, 50-kt and 64-kt winds based on a 5-year sample of operational track and intensity errors and climatological wind radii errors. Such products aid in operational assessment of the extent and/or arrival time of strong winds (e.g., gale force winds), which is very important for making pre-storm preparations by coastal residents, government agencies and other concerned parties. Also included in this group is the use of consensus and ensemble methods not only to determine a better deterministic forecast, but to convey a sense of certainty with that forecast. A couple methods, the FSU Superensemble (Mackey et al., 2005) and the STIPS consemble (Sampson et al. 2006), have shown some additional improvement in intensity forecasting. The second category of developing new technology in tropical cyclone forecasting concerns itself with the next generation of hurricane models. These models ideally will have their own initialization and data assimilation packages, include ocean and wave dynamics, and explicitly resolve convection. While the Geophysical Fluid Dynamics Laboratory (GFDL) has been the leader in the field and the current GFDL hurricane model is the operational state of the art. The GFDL model however is scheduled to be replaced in operations in 2007 by the Hurricane Weather Research and Forecast (HWRF) model, which National Centers for Environmental Prediction/Environmental Modeling Center (EMC) is actively developing the next generation hurricane modeling system. To accompany the HWRF development, EMC has developed new vortex initialization and data assimilation of real-time airborne Doppler data winds that will produce superior forecast of TC structure. In addition many GFDL studies have shown the positive impact of coupling the waves and the atmosphere on the hurricane structure forecasting. One of the most significant modeling challenges to improve numerical forecasts of hurricane structure and intensity in high-resolution hurricane models is the initialization of the hurricane vortex. In the initial implementation of HWRF in 2007, data assimilation will use EMC’s 3D variational analysis. To advance this effort in the HWRF, EMC is developing situation dependent background error (SDBE) covariances that will be incorporated into a local data assimilation scheme that will make use of real time Doppler radar data. It is widely recognized that the major outstanding analysis problem is improved formulation of the background error part of the analysis equation. Many improvements over the past 10 years have been in this area, including major upgrades to the ECMWF and NCEP systems. The SDBE approach attacks the fundamental analysis problem directly and is particularly relevant to the hurricane problem by capturing more of the hurricane structure through the flow dependent algorithms. The airborne Doppler radar from NOAA’s P-3’s and the newly funded instrument upgrade package on the NOAA

179

G-IV will provide hurricane core observations from the outflow layer to the surface to describe the three-dimensional wind structure of the storm for the data assimilation and improved vortex initialization. For storms approaching landfall, the data assimilation will also make use of the coastal WSR-88D high resolution radar data. It is anticipated that by 2010 the SDBE will be incorporated into a 4-D variational analysis scheme, which is under development at EMC. In previous versions of the GFDL hurricane model, the air-sea momentum flux was parameterized with a constant non-dimensional surface roughness regardless of wind speeds or sea states. This parameterization assumed a continual increase in Cd with wind speed. However, a number of studies (CBLAST, etc.) have suggested that the value of the drag coefficient and thus the Charnock coefficient (coefficient used by most MWP models to parameterize the boundary layer based on Monin-Obukhov similarity theory) depends on the sea state represented by the wave age. Lively debate is ongoing in the research community over this relationship. The major reason leading to the discrepancies among different studies is the paucity of in situ observations, especially in high wind speeds and young seas. The Charnock coefficient under hurricane conditions was also examined using a coupled wind-wave (CWW) model that includes the spectral peak in the surface wave directional frequency from WAVEWATCH III and a parameterized high frequency part of the spectrum in an updated version of the GFDL system (Falkovich 2005). The wave spectrum was introduced in the wave boundary layer model to estimate the Charnock coefficient at different wave evolution stages. It was found that the drag coefficient levels off at very high wind speeds, which is consistent with recent field observations (Powel 2003). The most important finding of this study is that the relationship between the Charnock coefficient and the input wave age (wave age determined by the peak frequency of wind energy input) is not unique, but strongly depends on wind speed. The regression lines between the input wave age and the Charnock coefficient have a negative slope at low wind speeds and a positive slope at high wind speeds (Moon et al 2004a; 2004b, 2004c). This behavior of the Charnock coefficient in high winds provides a plausible explanation why the drag coefficient under tropical cyclones, where seas tend to be extremely young, may be significantly reduced in high wind speeds. The above air-sea-wave coupling in the GFDL hurricane prediction system has shown very promising results on improving storm structure for Hurricane Ivan and TC’s from the 05 season. The coupling to the waves will become operational in the coupled HWRF system in 2007 and is expected to have significant impact on storm structure. Also included in the category of new technological developments are the new instruments and techniques needed not only to observe structure changes, but to develop a physical understanding of the important process related to tropical cyclone structure changes. A good example of this type of technology is the Stepped Frequency Microwave Radiometer (SFMR) (Uhlhorn et al. 2006), which when placed on a reconnaissance aircraft estimates the surface winds below the aircraft, in essence giving the forecaster and the researcher a two level analysis of the tropical cyclone wind structure.

c) Recommendations

After reviewing the state of tropical cyclone structure forecasting there are several recommendations that this working group has to make. The items range from operational instrumentation, to better use of existing technology. These are listed in numeric format.

1. There is a need for more operational scatterometery as it has become a vital tool for operational tropical cyclone forecasters and few future instruments are scheduled.

2. The development of consensus and ensemble based intensity forecast systems should be pursued to improve deterministic and probabilistic intensity prediction. These models/methods, including the STIPS consemble (Sampson et al 2006, cited 2006) and the FSU Superensemble forecast (Mackey et al., 2005), have out performed other intensity guidance.

180

3. At present none of the NWP guidance have skill in predicting wind radii, and NWP models that can predict both structure and intensity properly are likely several years away. In the mean time, some effort should be made to test whether output from existing NWP can be statistically fit to provide skillful guidance of tropical cyclone intensity and wind structure.

4. A concerted effort should be made to develop regional/specialized hurricane models that include specialized physical initialization and data assimilation packages.

5. New observational technology that benefits the tropical cyclone community should be made available to operational forecast centers.

Bibliography

Blackerby, J. S., 2005: Accuracy of western North Pacific tropical cyclone intensity guidance. M.S. thesis, Naval Postgraduate School, Monterey, CA, 107pp. Available at http://theses.nps.navy.mil/05Mar_Blackerby.pdf]

Booth, M. A., T. Lambert, J. Blackerby, and R. L. Elsberry, 2006: Accuracy of tropical cyclone intensity forecasts in the North Pacific and Atlantic. AMS 27th Conference on Hurricanes and Tropical Meteorology. 24-28 April, Monterey, CA. [available online at http://ams.confex.com/ams/pdfpapers/108805.pdf

Brown, D. P. and J. L. Franklin, 2004: Dvorak tropical cyclone wind speed biases determined from reconnaissance-based “best track” data (1997-2003). Preprints of the 26th Conference on Hurricanes and Tropical Meteorology, Miami, FL, 86-87.

____ and ____, 2002: Accuracy of pressure-wind relationships and Dvorak satellite intensity estimates for tropical cyclones determined from recent reconnaissance-bases “best track” data. Preprints of the 25th Conference on Hurricanes and Tropical Meteorology, San Diego, CA, 458-459.

Baik, J.-J., and J.-S. Paek, 2000: A neural network model for prediction typhoon intensity. J. Meteor. Soc. Japan, 78, 857-869.

Bender, M.A., and I. Ginis, 2000: Real-case simulations of hurricane-ocean interaction using a high-resolution coupled model: Effects on hurricane intensity. Mon. Wea. Rev., 128, 917-946.

Brueske, K. F., and C. S. Velden, 2003: Satellite –based tropical cyclone intensity estimation using he NOAA-KLM series Advanced Microwave Sounding Unit (AMSU). Mon. Wea. Rev., 131,687-697)

Cecil, D., T. A. Jones, and M. DeMaria, 2006: Passive-Microwave-Enhanced Statistical Hurricane Intensity Prediction Scheme. Wea. Forecasting, in press.

Chu, J-H., 1994: A regression model for the western North Pacific tropical cyclone intensity forecasts. NRL Memo. Rep. 7541-94-7215, Naval Research Laboratory, 33 pp. [Available from Naval Research Laboratory, 7 Grace Hopper Avenue, Monterey, CA 93943-5502]

Cocks, S. B. and W. M. Gray, 2002: Variability of the outer wind profiles of western North Pacific typhoons: Classifications and techniques for analysis and forecasting. Mon. Wea. Rev., 130, 1985-2005.

____, I. Johnson, R. Edson, M. A. Lander and C. P. Guard, 1999: Techniques for incorporating SSM/I imagery into Dvorak tropical cyclone intensity estimates. Preprints of the 23th Conference on Hurricanes and Tropical Meteorlogy, Dallas, TX, 584-587.

181

Cram, T., J.A. Knaff, and M. DeMaria, 2006: Objective Identification of Annular Hurricanes Using GOES and Reanalysis Data. AMS 27th Conference on Hurricanes and Tropical Meteorology. 24-28 April, Monterey, CA.

Cullen, M. J. P., 1993: The Unified Forecast/Climate Model. Meteor. Mag., 122, 81-122.

Davidson, N. E., and H. C. Weber, 2000: The BMRC high-resolution tropical cyclone prediction system: TC-LAPS. Mon. Wea. Rev., 128, 1245-1265.

Demuth, J. L. M. DeMaria, and J. A. Knaff, 2006: Improvement of advanced microwave sounding unit tropical cyclone intensity and size estimation algorithms. J. App. Met., in press.

____, M. DeMaria, J. A. Knaff, and T. H. Vonder Haar, 2004: Validation of an advanced microwave sounder unit (AMSU) tropical cyclone intensity and size estimation algorithm, J. App. Met., 43, 282-296.

DeMaria, M., and J. Kaplan, 1999: An updated statistical hurricane intensity prediction scheme (SHIPS) for the Atlantic and eastern north Pacific basins. Wea. Forecasting, 14, 326-337.

____, M. Mainelli, L.K. Shay, J.A. Knaff, J. Kaplan, 2005a: Further Improvement to the Statistical Hurricane Intensity Prediction Scheme (SHIPS). Wea. and Forecasting, 20: 4, 531–543.

_____, J.A. Knaff, K.J. Mueller, and R.M. Zehr, 2005b: Improvements in Deterministic and Probabilistic Tropical Cyclone Wind Predictions: A Joint Hurricane Testbed Project Update. 59th Interdepartmental Hurricane Conference, 7-11 March, Jacksonville, FL.

____, J. A. Knaff and J. Kaplan, 2006: On the Decay of Tropical Cyclone Winds Crossing Narrow Landmasses. J. App. Meteor.,45, 491–499.

Dvorak, V. F., 1975: Tropical cyclone intensity analysis and forecasting from satellite imagery. Mon. Wea. Rev., 103, 420-430.

____, 1984: Tropical cyclone intensity analysis using satellite data. NOAA Technical Report NESDIS 11, 45pp.

Ehlhorn, E. W., P. G. Black, J. L. Franklin, M. Goodberlet, J. Carswell, and A. S. Goldstein, 2006: Hurricane surface wind measurements from an operational stepped frequency microwave radiometer, Wea. Forecasting, in review.

Falkovich, A., I. Ginis, S. Lord, 2005: Implementation of data assimilation and ocean initialization for the coupled GFDL/URI hurricane prediction system. J. Atmos. and Ocean. Tech., 22, 1918–1932.

Franklin, J. L., cited 2006: 2005 National Hurricane Center forecast verification report. [available online at http://www.nhc.noaa.gov/verification/pdfs/Verification_2005.pdf]

Gallina M. G., and C.S. Velden, 2002: Environmental vertical wind shear and tropical cyclone intensity change utilizing enhanced satellite derived wind information. Preprints, 25th Conf. on Hurricanes and Tropical Meteorology, San Diego California, Amer. Meteor. Soc., 172.

Goerss, J. S., and R. A. Jeffries, 1994: Assimilation of synthetic tropical cyclone observations into the Navy Operational Global Atmospheric Prediction System. Wea. Forecasting, 9, 557-576.

182

Grell, G. A., J. Dudhia, and D. R. Stauffer, 1995: A description of the fifth-generation Penn State/NCAR Mesoscale Model (MM5). NCAR Tech. Note NCAR/TN-398 + STR, 122 pp.

Gross, J. M., M. DeMaria, J.A. Knaff, and C.R. Sampson, 2004: A New Method for Determining Tropical Cyclone Wind Forecast Probabilities. AMS 26th Conference on Hurricanes and Tropical Meteorology. 3-7 May, Miami, FL.

Heming, J. T., J. C. L. Chan, and A. M. Radford, 1995: A new scheme for the initialization of tropical cyclones in the UK Meteorological Office global model. Meteor. Appl., 2, 171-184.

Hodur, R. M., 1997: The Naval Research Laboratory's Coupled Ocean/Atmosphere Mesoscale Prediction System (COAMPS). Mon. Wea. Rev., 125, 1414-1430.

Hogan, T., and T. Rosmond. 1991: The description of the Navy Operational Global Atmospheric Prediction System’s spectral forecast model. Mon. Wea. Rev., 119, 1786-1815.

Jarvinen, B. R. and C. J. Neumann, 1979: Statistical forecasts of tropical cyclone intensity for the North Atlantic basin. NOAA Tech. Memo. NWS NHC –10. 22 pp. [Available from NTIS, Technology Administration, U.S. Dept. of Commerce, Springfield, VA 22161]

JMA, cited 2006: Annual Report on Activities of the RSMC Tokyo-Typhoon Center. [available online at http://www.jma.go.jp/jma/jma-eng/jma-center/rsmc-hp-pub-eg/annualreport.html]

Kaplan, J., and M. DeMaria, 1995: A simple empirical model for predicting the decay of tropical cyclone winds after landfall. J. Appl. Meteor., 34, 2499-2512.

_____ and _____, 2001: On the decay of tropical cyclone winds after landfall in the New England area. J. Appl. Meteor., 40, 280-286.

_____ and _____, 2003: Large-Scale Characteristics of Rapidly Intensifying Tropical Cyclones in the North Atlantic Basin. Wea. Forecasting, 18, 1093–1108.

Knabb R., E. Rappaport, M. Mainelli, J. Franklin, C. Lauer and Alison Krautkramer, cited 2006: Progress toward operational implementation of tropical cyclone wind probability products. [available online at http://www.ofcm.gov/ihc06/Presentations/06%20session6%20Decision-making%20Products/s6-04Session%206%20(knabb).ppt#492,31,Slide 31 ]

Knaff J. A., M. DeMaria, C. R. Sampson, J.M. Gross, 2003: Statistical, five-day tropical cyclone intensity forecasts derived from climatology and persistence. Wea. Forecasting, 18, 80-92.

_____, J.P. Kossin, and M. DeMaria, 2003: Annular Hurricanes. Wea. Forecasting, 18:2, 204-223.

_____, C.R. Sampson, M. DeMaria, 2005: An operational statistical typhoon intensity prediction scheme for the Western North Pacific. Wea. Forecasting, 20, 688–699.

Kossin, J. P., J. A. Knaff, H. I. Berger, D. C. Herndon, T. A. Cram, C. S. Velden, R. J. Murnane, and J. D. Hawkins, 2006: Estimating hurricane wind structure in the absence of aircraft reconnaissance. Wea. Forecasting, in press.

____, H. Berger, J. Hawkins, J. Knaff, M. DeMaria, and T. Cram, cited 2006: Development of a

183

secondary eyewall formation index for improvement of tropical cyclone intensity forecasting , Presentation at the 60th Interdepartmental Hurricane Conference, Mobile Al. [available at http://www.ofcm.gov/ihc06/Presentations/02%20session2%20Observing%20the%20TC/s2-11kossin.ppt ]

Kuma, K., 1996: NWP activities at Japan Meteorological Agency. Preprints, 11th Conf. on Numerical Weather Prediction, Norfolk, VA, Amer. Meteor. Soc., J15-J16.

Kurihara, Y., M. A. Bender, and R. J. Ross, 1993: An initialization scheme of hurricane models by vortex specification. Mon. Wea. Rev., 121, 2030-2045.

______, _______, R. E. Tuleya, and R. J. Ross, 1995: Improvements in the GFDL hurricane prediction system. Mon. Wea. Rev., 123, 2791-2801.

_______, R. E. Tuleya, and M. A. Bender, 1998: The GFDL hurricane prediction system and its performance in the 1995 hurricane season. Mon. Wea. Rev., 126, 1306-1322.

Lambert, T. D. B., 2005: Accuracy of Atlantic and Eastern North Pacific tropical cyclone intensity guidance. M.S. thesis, Naval Postgraduate School, Monterey, CA, 119pp. [Available at http://theses.nps.navy.mil/05Sep_Lambert.pdf]

Lord, S. J., 1993: Recent developments in tropical cyclone track forecasting with the NMC global analysis and forecast system. Preprints, 20th Conf. on Hurricanes and Tropical Meteorology, San Antonio, TX, Amer. Meteor. Soc., 290-291.

Mackey, B.P., M.K. Biswas, T.N. Krishnamurti, 2005: Performance of the Florida State University Superensemble during 2004, Presentation at the 59th Interdepartmental Hurricane Conference, Jacksonville, FL. [available at http://www.ofcm.gov/ihc05/Presentations/02%20session2/s2-10mackey.ppt]

Martin, J. D. 1988: Tropical Cyclone Observation and Forecasting with and without Aircraft Reconnaissance, Department of Atmospheric Science Paper #428 , 114 pp. [Available from Colorado State University, Fort Collins, CO 80523]

McAdie, C. J., 2004: Development of a wind-radii CLIPER model. Preprints of the 24th Conference on Hurricanes and Tropical Meteorology, Miami, FL, 170-171.

Merrill R. T., 1984: A comparison of large and small tropical cyclones. Mon. Wea. Rev., 112, 1408–1418.

Moon, I.-J., I. Ginis, and T. Hara, 2004a: Effect of surface waves on Charnock coefficient under tropical cyclones, Geophys. Res. Lett. 31, L20302.

____, T. Hara, I. Ginis et al., 2004b: Effect of surface waves on air-sea momentum exchange. Part I: Effect of mature and growing seas, J. Atmos. Sci., 61, 2321– 2333.

____, I. Ginis, and T. Hara, 2004c: Effect of surface waves on air-sea momentum exchange. Part II: Behavior of drag coefficient under tropical cyclones, J. Atmos. Sci., 61, 2334– 2348.

Mueller, K. J., M. DeMaria, J. A. Knaff, J. P. Kossin, and T. H. Vonder Haar, 2006: Objective Estimation of Tropical Cyclone Wind Structure from Infrared Satellite Data, Wea. Forecasting (in press)

184

Olander, T. L, and C. S. Velden, 2006: The Advance Dvorak Technique (ADT) –continued development of an objective scheme to estimate tropical cyclone intensity using geostationary infrared satellite imagery. Wea. Forecasting, in press.

Powell, M. D., P. J. Vickery, and T. A. Reinhold (2003), Reduced drag coefficient for high wind speeds in tropical cyclones, Nature, 422, 279– 283.

Rennick, M. A., 1999: Performance of the Navy's tropical cyclone prediction model in the western North Pacific basin during 1996. Wea. Forecasting, 14, 3-14.

Sampson, C. R., and A. J. Schrader, 2000: The Automated Tropical Cyclone Forecasting System (Version 3.2). Bull. Amer. Meteor. Soc., 81, 1131-1240.

______, R. J. Miller, R. A. Kreitner, and T. L. Tsui, 1990: Tropical cyclone track objective aids for the microcomputer: PCLM, XTRP, PCHP. Naval Oceanographic and Atmospheric Research Laboratory, Tech Note 61, 15pp. [Available from Naval Research Laboratory, 7 Grace Hopper Avenue, Monterey, CA 93943-5502]

______,J. A. Knaff, M. DeMaria, cited 2006, recorded presentation of “A statistical intensity model consensus for the Joint Typhoon Warning Center.” AMS 27th Conference on Hurricanes and Tropical Meteorology. 24-28 April, Monterey, CA. [available online at http://ams.confex.com/webex/ams/27Hurricanes/107554/26f8ed850faf711503a0a3c1af35bc16/107554.wrf

______, J.A. Knaff, and M. DeMaria , 2006:

Velden, C. S., T. L. Olander and R. M. Zehr, 1998: Development of an objective scheme to estimate tropical cyclone intensity from digital geostationalry satellite infrared imagery. Weat. Forecating, 13, 172-186.

Weatherford C. L., and W. M. Gray, 1988: Typhoon structure as revealed by aircraft reconnaissance. Part II: Structural variability. Mon. Wea. Rev., 116, 1044-1056.

185

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES Topic 1a : Special focus session on: "Tutorial on the use of satellite data to define tropical cyclone structure". Rapporteur: C.S. Velden University of Wisconsin - CIMSS 1225 West Dayton St. Madison WI 53706, USA E-mail: chrisv@ssec.wisc.edu Fax: 608 262 5974 Working Group: J. Hawkins, R. Zehr, R. Edson, D. Herndon, J. Kossin, T. Nakazawa, P. Caroff Abstract: The aim here is to examine the advances that have occurred in satellite-based remote sensing since the IWTC-V, and to break new ground in some areas. In this report we briefly summarize new satellite-based methods to estimate the intensity and structure of TCs. The report is broken down into two main areas: Applications from IR images, and microwave data. After some background in each area, the focus is on research algorithms and potential operational applications that have been developed since the last IWTC 4 years ago. 1.a.1 IR Applications Early applications of IR satellite images for TCs were for tracking (center fixing) and intensity following the Dvorak technique (Dvorak 1975; Dvorak 1984;Velden et al, 2006). Those type applications remain today as the primary and most important applications. Since the original Dvorak development work, results have likely improved due to improved enhancements, animations, better image navigation, higher spatial resolution, and more frequent temporal resolution. 1.a.2. Objective and Advanced Dvorak Technique (ODT, AODT, ADT) Important applications followed Dvorak (1984) with the improvement of the original Dvorak algorithm (Zehr, 1989) and the Objective Dvorak Technique (ODT) Velden et al 1998). The original goal of the ODT was to achieve the accuracy of the subjective Dvorak operational method using computer-based, objective methodology. This goal was accomplished, however, important limitations still existed. The ODT only operated on storms that possessed an intensity at, or greater than, minimal hurricane/ typhoon strength. Also, the ODT algorithm still required the manual selection of the storm center location needed for the subsequent analysis. These issues motivated the development of the Advanced Objective Dvorak Technique (AODT) (Olander and Velden 2004). The most recent version of the objective Dvorak algorithm progression is the Advanced Dvorak Technique (ADT) (Olander and Velden 2006). Unlike the ODT and AODT that centered on attempts to mimic the subjective technique, the ADT research efforts have focused on revising some of the digital IR thresholds and rules, and extending the method beyond the original application and constraints. The ADT is fully automated and providing forecasters with an objective tool as guidance in their real time TC analysis, and as a comparison to their subjective Dvorak estimates.

186

1.a.3. Other IR applications/products a. Cold IR cloud area time series. This product is displayed as time series plots of percent coverage of pixels colder than an IR temperature threshold within a circle of radius IR. Typical R used is 2- 4 deg lat (222-444 km), with typical thresholds of -50 to -75C. Zehr has more recently compiled values at time of maximum intensity for all 45 Atlantic major hurricanes during 1995-2005 (Fig.1.a.1).

Figure 1.a.1. Percent coverage of IR pixels <-60C at R=0-444 km at maximum intensity, 1995-2005 Atlantic intense hurricanes.

b. Azimuthal mean time series plots. This product displays time series of radial profiles of

azimuthal averages of IR temperature, and was developed by Kossin (2002). c. IR asymmetry computations. Zehr’s (2003) case study on Hurricane Bertha shows that the bearing and distance from TC center of cold IR cloud centroids are related to the environmental vertical wind shear (Fig. 1.a.2).

187

Figure 1.a.2. Centroid locations of IR cloud areas with Hurricane Georges, 23 UTC 17 Sep 98. d. Center relative IR average images. Animations remove motion and short time scale variability from the IR images. 6-h averages at 3-h intervals have been used recently with Hurricane Wilma. ftp://rammftp.cira.colostate.edu/Zehr/IWTC06/Wilma-SR-AV.avi e. Inclusion of IR data into statistical forecast models. DeMaria et al (2005) have shown that inclusion of GOES quantitative IR derived data have produced small improvements in the SHIPS forecasts. The GOES IR data significantly improved the east Pacific forecasts by up to 7% at 12–72 h. The GOES predictors are 1) the percent of the area (pixel count) from 50 to 200 km from the storm center where TB is colder than −20°C and 2) the standard deviation of TB (relative to the azimuthal average) averaged from 100 to 300 km. f. RAMM/CIRA Tropical Cyclone IR archive: In 1998, RAMM/CIRA began archiving IR images at 4 km Mercator resolution covering the entire life cycle of Atlantic and eastern Pacific tropical cyclones (Zehr, 2000). The archive begins in 1995 for Atlantic hurricanes, and in 2004 it was automated with global coverage. The GOES images are at 30-min interval. The archive through June 2006, includes about 475 tropical cyclones. Using the RAMM/CIRA IR archive, Kossin (2002) has documented diurnal and semidiurnal IR variability. Knaff et al (2003) have analyzed a subset of tropical cyclones, called annular hurricanes. g. High resolution IR images. 1-km resolution IR images from the AVHRR on NOAA satellites and the OLS on DMSP have been available for many years, but not with global coverage. The MODIS on NASA Terra and Aqua now provide 1-km resolution globally. 1-km IR images reveal features such as cyclonically curved thin cold cloud lines and transverse bands at hurricane cloud top that are not well observed by the lower resolution GOES (Fig. 1.a.3).

188

Figure 1.a.3. 1-km IR images of Hurricane Isabel, 10-12 September, 2003, from both MODIS and

NOAA AVHRR. h. IR relationships with wind radii and TC structure: Kossin et al (2006a) have applied IR data to new objective methods of estimating radius of maximum wind (RMW), along with the standard operational wind radii analyses and forecasts (R-35, R_50, R-64). Another routine was developed which generates the entire 2-dimensional wind field within 200 km radius. This provides inner-core winds, which has historically been limited to cases where aircraft reconnaissance is available. These inner core winds can be combined with winds from other satellite platforms to form complete wind fields in basins where no aircraft reconnaissance is available. i. IR relationships with wind radii and TC structure: Mueller et al (2006) use aircraft observations along with statistical relationships with IR data to estimate radius of maximum wind and TC structure. j. Objective IR identification of annular hurricanes: Cram et al (2006) have derived an algorithm that uses IR data to objectively identify annular hurricanes. The algorithm is based on linear discriminant analysis, and is being combined with a similar algorithm being developed at CIMSS. k. By enhancing the signal to noise ratio using Principal Component Analysis, Kossin et al. (2006b) found that IR imagery does contain information about the onset of eyewall replacement cycles. This information was combined with other information from microwave imagery and environmental fields to form an objective index to calculate the probability of secondary eyewall formation. The method uses linear discriminant analysis and Bayesian Classification to provide forecasters with a realtime tool.

189

1.a.4. Microwave Sensors/Applications a. Imagers Passive microwave imagery (PMW) applications have progressed due to the recent launch of two new microwave imagers and the extensions of existing programs, especially the Tropical Rainfall Measuring Mission (TRMM) till at least 2009. Passive microwave imagers map key TC structural rainband organization and eyewall development by “seeing through” non-raining clouds. Similar details can’t be routinely extracted in visible/Infrared (vis/IR) data due to upper-level cloud obscuration (Hawkins, et al., 2001, and Lee, et al., 2002). Figure 1.a.4 illustrates a classic case where clouds hide the true TC low-level center that can be readily identified in microwave imagery products.

Figure 1.a.4: Multiple typhoon Mindulle views with GMS-5 IR (left, indicating circulation center on left-center of picture) and NRL SSM/I 85 GHz “composite” product (right, clearly showing low-level center well right of main convection (red)). Access to global near real-time PMW imagery covering all active TCs can be found on two US Navy web sites: http://www.nrlmry.navy.mil/tc_pages/tc_home.html http://152.80.49.216/tc-bin/tc_home.cgi The WindSat polarimetric radiometer is a US Navy research sensor launched onboard the Coriolis spacecraft in January 2003 to test the extraction of surface wind vectors via a passive microwave radiometer (Gaiser, et al., 2004). The large antenna provides excellent spatial resolution 37 GHz brightness temperatures (Tb) across the 1025-km swath and permits users to monitor TC structure (eyewall configuration changes), similar to capabilities in existing TMI and AMSR-E data (Hawkins, et al., 2006). Thus, although WindSat does not have a scattering channel (85-91 GHz), it can significantly contribute to the overall mapping goal and is operationally used by numerous agencies to create TC center fixes. 37 GHz data provides a lower-level view of TC structure than 85-91 GHz data and often provides a less ambiguous center fix (confusion can arise concerning low-level and mid-level circulation centers). However, the poor spatial resolution 37 GHz data on the SSM/I and SSMIS severely limit this channel as clearly depicted in Fig. 1.a.5. Therefore, enhanced temporal sampling of high-resolution 37 GHz data will permit improved satellite reconnaissance world-wide during WindSat’s mission.

190

Figure 1.a.5: Comparison of 37 GHz Tbs for SSM/I (left side) and WindSat (right) for typhoon Songda in the western Pacific on Sept 4, 2004. Note the inability to observe inner core structure in poor resolution SSM/I data versus the double eyes in WindSat. The operational Special Sensor Microwave Imager Sounder (SSMIS) was launched in October 2003 onboard a Defense Meteorological Satellite Program (DMSP F-16) spacecraft (Wessel, et al., 2004). The SSMIS is the follow-on to the Special Sensor Microwave/Imager (SSM/I) with a larger swath (1700 versus 1400-km) and four more (F-17, 18, 19, 20) will be orbited over the next 1-6 years. The imager channels are essentially unchanged from the SSM/I (91 GHz versus 85 GHz), but it also includes collocated sounder channels that have multiple advantages. SSMIS sensors will replace the older SSM/I data and already aid TC monitoring globally (Fig. 1.a.6).

Figure 1.a.6: Sequence of three SSMIS 91 GHz H-pol Tbs of Hurricane Rita while in the Gulf of Mexico. F-16 SSMIS data helped provide temporal sampling to monitor the storms eyewall cycle evolution. Note double eyewalls in all three snapshots. Mitigating the PMW temporal sampling issue can also be accomplished by incorporating methods to “bridge-the-gap” between two successive microwave images. Wimmers and Velden, 2004 highlight a technique to produce near real-time microwave animations that incorporate image “morphing” tools to visually illustrate how storm features evolve from one image to another, typically within 2-7 hours of one another as noted in Figure 1.a.7. The morphing often lets the user better understand eyewall replacement cycles (the change from one to two and back to one eyewall for intense TCs) that can be

191

difficult to discern in single images. Near real-time morphing demonstrations are available from the CIMSS website in addition to more details explaining the methodology: http://cimss.ssec.wisc.edu/tropic/real-time/marti/marti.html

Figure 1.a.7: Example of Hurricane Ivan 85-89 GHz microwave images available to the “morphing” technique that will then attempt to fill in the gaps between successive images from the various sensors. The top scale is “hours” during Sept. 12-13, 2004. No additional microwave scatterometers have been added to the TC observing system, but fortunately still include the QuikSCAT scatterometer launched in 1999. QuikSCAT’s 1800-km swath has proven to be beneficial in mapping TC wind fields and was shown in one study to significantly enhance our ability to monitor TC genesis (Sharp, et al, 2002) by mapping vorticity. The ocean surface wind vectors have problems in the tropical cyclone inner-core region due to both rain attenuation and the storm’s inherently strong wind speed and directional gradients. Several methods now exist to extract wind vectors at “high” resolution utilizing special processing (Halterman and Long, 2006). Using microwave radiometer data, Cecil and Zipser (1999) investigated the relationship between polarization corrected temperature (PCT) of 85 GHz channel and present and future maximum wind speed and tendency of TC intensity. Recently Hoshino and Nakazawa(2006) have proposed a new estimation method of TC utilizing 10, 19, 21, 37 and 85 GHz channel TRMM Microwave Imager (TMI) data. In contrast to the previous studies, Hoshino and Nakazawa(2006) found that the parameters with lower frequency channels of 10 or 19 GHz give higher correlation. The highest correlation coefficient obtained is 0.7 and the root mean square error (RMSE) of the regression between a parameter of highest correlation case is found to be 6 m/s. b. Tropical Cyclone Intensity Estimation Using Microwave Sounders The AMSU instrument is a cross-track scanning microwave sounder containing 15 channels with frequencies ranging from 23.8 – 89 GHz. Resolution varies from 48-km at nadir to ~ 100-km near the limb. The series began in 1998 aboard NOAA-15 and now include three working sensors aboard NOAA polar orbiters (NOAA-15, 16, and 18) and an AMSU is also flown aboard NASA’s Aqua satellite. The primary channels used for TC monitoring are the ~55 GHz oxygen band channels 5-8 and the moisture channels 1-4 (AMSU-A) and 16 (AMSU-B). Previous microwave sounders (MSU) had limited TC applications due to poor spatial resolution, but AMSU’s improvements have enabled the development of operational algorithms for TC intensity and structure monitoring.

06 08 10 12 14 16 18 20 22 00Sept 12 Sept 13

AMSR-E TRMM TMI SSMI SSMI

06 08 10 12 14 16 18 20 22 00Sept 12 Sept 13

AMSR-E TRMM TMI SSMI SSMI

06 08 10 12 14 16 18 20 22 00Sept 12 Sept 13

AMSR-E TRMM TMI SSMI SSMI

06 08 10 12 14 16 18 20 22 00Sept 12 Sept 13

AMSR-E TRMM TMI SSMI SSMI

192

AMSU-A channels 5-8 have weighting functions located in the mid to upper-troposphere and are well-suited to monitor TC warm core temperature structure. The magnitude of the TC warm core is directly correlated with TC intensity, thus AMSU and Tb anomalies can be used to estimate TC intensity with skill that is equivalent or superior to the IR-based Dvorak method (see Fig. 1.a.8).

Figure 1.a.8. Relationship between AMSU-A measured brightness temperature anomaly for channel 8 (Tb, K) and TC pressure anomaly (environmental pressure – minimum sea level pressure [MSLP], hPa) using data from 1998-2005 (~700 observations). Potential AMSU limitations: 1) The AMSU spatial resolution is insufficient to completely measure TC warm-core features. The average TC eye diameter is 45-km with some storms exhibiting eyes much smaller, meaning that AMSU (48-km at nadir) will sub-sample most storms. 2) Hydrometeor effects: The presence of large quantities of hydrometeors acts to decrease the satellite observed Tb, reducing warm core measurements. 3) Vertical Resolution: The true warm core altitude may fall in between AMSU channels and thus the instrument may underestimate the maximum warm core signal. 4) The TC center may horizontally fall between AMSU Field of View (FOV) positions. TC eye diameters < 20-km may result in cases where problems 1-4 occur simultaneously with much of the eyewall being located within the AMSU FOV.

Horizontal resolution is the most important of the four major error sources. Intensity estimate errors of more than 40 hPa are possible for storms with very small pinhole eyes (< 10-km such as Hurricane Wilma).

There are currently two operational AMSU-based intensity algorithms. CIRA has developed an algorithm that attempts to retrieve the vertical temperature structure (Fig. 1.a.9, left) that is then combined with a number of additional AMSU-based predictors to derive a MSLP and maximum sustained wind (MSW) estimate. The CIMSS algorithm uses the raw Tb anomalies (Fig. 1.a.9, right) which have been matched to MSLP observations along with AMSU FOV and TC structure information to estimate MSLP.

193

Figure 1.a.9. Representation of AMSU-A retrieved vertical Tb anomaly from CIRA (left) and horizontal raw Tb anomaly for channel 8 from CIMSS (right) for Hurricane Floyd on September 14, 1999. Limb corrections are applied to AMSU data to compensate for decreasing Tbs as one moves away from nadir. A data set including 90 AMSU TC overpasses (Atlantic) was also created to help remove biases resulting from poor AMSU eye resolution (Herndon, 2004). All eye sizes < 50-km and coincident with reconnaissance eye reports from 1998-2002, radar and microwave imagery (AMSR-E, TMI, SSM/I) were compiled to help remove any systematic biases. Figure 1.a.10 highlights the problems between sensor footprint and eye diameter for Hurricane Michelle. Eye size is derived by the ADT algorithm for “clear eye” cases and from the radius of maximum winds (RMW) entered into ATCF. An initial estimate of MSLP is performed using the raw Tb anomaly. If the eye size is small compared to the AMSU FOV then a bias correction is subtracted. The Tb signal can also change slightly from one estimate to the next as a function of storm position within the scan swath. To account for these changes, a second regression predictor (distance in FOV steps from nadir) was added. The end result is slight intensity increases for cases near the swath edge and slight decreases for those cases near nadir. Improved performance was validated and the modification was adopted for real-time use in 2003.

194

Figure 1.a.10. AMSU FOV example for Hurricane Michelle illustrating non-resolved (top) and resolved (bottom) eye size cases (right side) and resulting bias corrections (left panel). Bias correction is applied when eye size is smaller than AMSU FOV. Because not all storms form in the same ambient pressure environment the algorithm was refined to estimate MSLP anomaly instead of MSLP. The outer closed isobar from ATCF is used as the environmental pressure. The CIMSS AMSU algorithm uses pressure units for intensity estimates since we have a higher confidence with respect to accuracy of the MSLP training data than the MSW data. More recently, a conversion to wind estimates has been included at the request of TC warning centers. Initially the wind estimate was derived from a simple pressure/wind relationship developed by Kraft for Atlantic TC’s. Because this method used del-P, it was also easily used in other basins where the ambient pressure environment is often lower. However, this method has proved unsatisfying since a simple pressure/wind relationship is insufficient for all cases. Improved wind estimates will be derived directly by comparing the MSW to Tb anomalies as is done for the pressure estimate. Further adjustments are planned based on measures of convective vigor from available IR and microwave channels. Hydrometeor contamination problems decrease effective Tbs, requiring a correction using AMSU-A channels 2 and 15. Figure 1.a.11 is an example of this effect. A two-channel difference can determine if Tbs are being depressed due to hydrometeor scattering. A correction is then applied to the raw Tbs resulting in a new set of regression coefficients (Wacker et al, 2004).

195

Figure 1.a.11. Channel 7 Tb (K) showing hydrometeor contamination (left, with artificial bullseye) and Tb map after rain corrections (right) mitigate rain impact. The CIRA algorithm begins with a statistical temperature retrieval (Goldberg et al 2001) to estimate the true atmospheric temperature from the AMSU limb-corrected Tb at 40 vertical levels. Of these 40 levels, 23 are used in the TC intensity algorithm. AMSU-derived cloud liquid water (CLW) values are used to correct the temperature profiles and remove the effects of scattering due to liquid water. A second correction is then applied to account for the effects of ice scattering. The hydrometeor corrected temperatures are then interpolated to a radial grid and azimuthally averaged from the estimated TC center (interpolated from TC warnings) out to a radius of 600-km. Surface temperature and surface pressure boundary conditions are supplied by the NCEP global analysis data. Geopotential heights are then derived as a function of pressure using the hydrostatic equation from 50 hPa to 920 hPa. With the height and temperature data for all levels above the surface the hydrostatic equation is then integrated downward to obtain the surface pressure field within the domain. The original CIRA TC intensity algorithm used 20 potential estimators to estimate TC MSLP and MSW (Demuth et al, 2004). In addition the algorithm provides information about TC wind structure with estimates of 34, 50 and 65 knot wind radii. In 2005 four additional estimators were added: (Demuth et al, 2006): TMAX2 The squared value of the maximum temperature perturbation CLWAVE2 Squared value of CLW content (CLWAVE) TMAXxCLWAVE2 TMAX multiplied by CLWAVE2 P600* Surface pressure (hPa) at r = 600 km *These values are not derived from the AMSU data. The above parameters were matched to more than 2600 MSW and MSLP values from Best Track data covering TC’s in the Western Pacific, Indian Ocean and Southern Hemisphere from 2002-2004 and the Atlantic, East Pacific and Central Pacific from 1999-2004. The resulting multiple regression forms the basis of the algorithm to predict MSLP and MSW for storms within 700 km of the AMSU sub-point. A similar statistical approach using sub-sets of the same parameters is used to estimate the 34, 50 and 65 kt wind radii. However in order to improve the wind radii estimates only cases in which reconnaissance data was available within 12 hours of the AMSU pass time were used for the radii

196

algorithm development (N= 255, 170 and 120 for the 34, 50 and 65 kt radii respectively). Wind radii estimates are produced for all 4 quadrants of the TC allowing an estimate of wind asymmetry. Tropical cyclone eyewall replacement cycles (ERCs) are associated with significant, and often rapid, intensity changes, but we presently have no algorithms in place to diagnose their onset. The inception of an ERC can be identified by secondary eyewall formation (SEF) as outer rainbands begin to encircle the existing single eyewall. Using official intensity forecast errors validated against aircraft reconnaissance, we find that stronger storms, such as those that undergo ERCs, are typically under-forecast. In cases with Vmax > 100 kt, 24-hr intensity forecasts have a bias of –1.4 kt. In comparison, near the time of SEF, 24-hr forecast bias increases to +2.4 kt and intensity is systematically over-forecast (Kossin et al, 2006). The mean absolute 24-hr forecast error near the time of SEF increases by 20% since forecasters are unable to anticipate SEF processes that typically create short-term weakening. Therefore, accurate intensity forecasts will depend in part on our ability to diagnose and forecast SEF. Ideally, SEF monitoring via geostationary data is highly preferred due to superb temporal sampling, but vis/IR data is severely challenged by obscuring cirrus residing above the strong storms undergoing SEF. Upper-level cirrus masks the convective spiral band structure and makes it difficult to observe the symmetric organization of this outer convection into a secondary eyewall. A solution to this challenge is the application of microwave satellite data, which penetrates non-raining clouds and enables analysts to monitor the outer band convective evolution. SEF signals have been identified in microwave satellite imagery and an objective SEF-index has been created that can be applied operationally. The index gives a “YES/NO” classification based on Linear Discriminant Functions, and a probability of YES/NO using Bayesian Classification with K-Nearest Neighbor probability density functions. The algorithm is being combined with a similar index that diagnoses the formation of annular hurricanes, and is being constructed by members of the NOAA/NESDIS RAMMB team. Since annular hurricanes tend to maintain a more steady intensity than average while they are annular, they represent a very different population than storms that undergo ERCs and exhibit more intensity variance than average during the ERC. An index that can diagnose both SEF and annular hurricane formation should serve as a useful modifier of intensity change predicted by numerical or statistical guidance. c. Rainrate Applications A recent NOAA/AOML/HRD effort has filled an important gap by formulating a TC rainfall climatology via TMI data referred to as R-CLIPER. Lonfat and Marks, 2004, incorporated three years (1998-2000) of global TMI data to map rainfall in 260 TCs. The high resolution TMI data captures much of the TC rainfall events and is superior to coarser data sets such as SSM/I, SSMIS, and AMSU-B. TMI rain data was azimuthally averaged and binned by storm basin and intensity to permit a climatological profile of rain intensity radially outward from storm center. The National Hurricane Center incorporates R-CLIPER and official track forecasts to create a graphical 24 and 72-hr rain accumulation forecast for all landfalling storms (see Fig. 1.a.12 for example with Hurricane Frances). The products are a starting basis forecasters can use and modify depending on specific storm characteristics (e.g., bigger than normal for Cat 2 storm).

197

Figure 1.a.12. Example R-CLIPER rain accumulation (inches) graphic depicting forecast rainfall totals for Hurricane Jeanine during a Florida landfall in Sept. 2004. Available in near real-time at NHC, Miami. NOAA’s Tropical Rainfall Potential (TRaP) technique generates 24-hour rainfall forecasts using a combination of microwave derived rainrates and official TC track forecasts (Ferraro, et al., 2005). Microwave overpasses produce a snapshot of rainfall that is then advected along the forecast track output from the National Hurricane Center (NHC, Miami) or the Joint Typhoon Warning Center (JTWC, Pearl Harbor, HI). Although the methodology does not take into account storm rotation and growing/decaying rainbands and cells, it produces ballpark values that can assist forecasters and emergency managers. As one might expect, best results occur when the highest spatial resolution imager (TMI) served as the base rainfall map. Products using SSM/I data and its inherently poorer spatial resolution faired worse. Passive microwave imagers have been the focus of numerous efforts to create physically-based rainrates that are inherently more accurate than IR-inferred estimates. Thus, the satellite rainfall algorithm community has trended towards “merged” techniques that capitalize on the accuracy of PMW and temporal sampling of GEO IR. Ready access to global passive microwave data in near real-time (1-4 hours) and digital geostationary data sets now enable multiple merged algorithms for operations and research (Turk and Bauer, 2005). NOAA creates the CMORPH product that uses geostationary cloud motions to advect microwave-derived precipitation forward in time (Joyce et al., 2004). NRL brings together rainrates from ten microwave imagers/sounders to dynamically update the relationship between IR-based cloud top temperatures and rainrate (Turk and Miller, 2005). Huffman, et al, 2003 incorporate a slightly different methodology to create microwave-IR correlations via a Multi-satellite Precipitation Analysis (MPA). These rain monitoring efforts are geared toward global applications that encompass a true diversity in precipitation regimes, but can be applied to TCs. In addition, a new inversion-based algorithm at NASA GSFC using both TRMM TMI and PR data is under development. Vertical profiles of precipitation ice water and liquid water content are created via these merged data sets and validation statistics are quite promising (Jiang and Zipser, 2006). The ability to take advantage of combined TMI/PR data and 3-D information is critical to algorithm advancement. d. Scatterometer Data Applications Since the FIFTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES in Cairns, Australia, December 2002, there has been a substantial increase in the world-wide use of the QuikSCAT

198

scatterometer to aid in many basic forecasting requirements over the ocean. With this has come a better understanding of certain meteorological and oceanographic events that up to now have not been adequately observed over these data sparse regions. Examples are plenty, including improvements in marine ocean surface applications such as with ice movement and predicting accurate sea heights and swell. Numerical models (NWP) have improved with the better understanding of the marine boundary layer and more extensive surface data assimilation. The occurrence and timing of extreme ocean weather events such as the persistence of large areas of hurricane-force winds were previously not well understood until they were more accurately witnessed with the QuikSCAT data. And for almost all aspects of the tropical cyclone warning process (genesis, positioning, structure, intensity, extratropical transition/dissipation), the QuikSCAT data have been at the forefront of providing the tropical cyclone forecaster and researcher a better understanding of the many tropical cyclone processes. One reason why the QuikSCAT scatterometer data have been such a success is because the forecaster and analyst have gained a better understanding of how to use the data and how to better interpret some of the known problem areas such as its range of accurate wind speeds, the ambiguity selection process and the effects of rain. This knowledge has also helped the developers of QuikSCAT improve the product as well. Of key importance is the need to integrate all available satellite-based remote sensing data into a combined analysis. In this way, each additional source of data can complement the other and lessen some of the problem areas. Another important development is through the sophisticated methods currently being used to obtain special high resolution data via the processing of the pulse information in the QuikSCAT source. In this section some of the new techniques used in the analysis of tropical cyclones are demonstrated and hints at how this may help to gain a better knowledge of the tropical cyclone structure are presented. First a recap of the sources and types of data available for near real-time use is given. QuikSCAT Data Sources and Data Types: The QuikSCAT satellite is a polar-orbiting, sun-synchronous satellite with an equator-ascending time of approximately 0600 local (+/- 30 minutes) (Fig 1.a.13). Wind vector solutions and up to four ambiguity solutions are stored in 25 km X 25 km wind vector cells (wvc) over an 1800 km wide swath. Besides including the position of the center of each cell, each wvc also contains various flagging indicators that include possible contamination with rain, land, and ice.

Fig 1.a.13: Examples of daily QuikSCAT Near-Real Time (NRT) passes which are generally available within 2 ½ hours of data time from the NOAA/NESDIS site. Daily Global Views: There are two primary US sources of near real-time global QuikSCAT scatterometer data available to the general public: the US Navy’s FNMOC (public) site and the

199

NOAA/NESDIS site (Fig 1.a.14). In addition, both sites offer selective enhanced views for active tropical cyclone warnings and INVEST areas. The global FNMOC display offers two variations of selected wind vectors for display: a Near-Real Time (NRT)-NOAA/GFS Analysis (GFS) version (identical to the data displayed by NOAA/NESDIS) and a NRT-Navy NOGAPS (NOGAPS) version. Both versions originate from the same QuikSCAT1 wind retrieval algorithm, but use a different Numerical Weather Prediction (NWP) model (GFS versus NOGAPS) to help in the ambiguity selection process (the method of determining which of up to four (1-4) solutions for any point is the most likely choice). The global NOAA/NESDIS NRT-GFS version is identical to the FNMOC NRT-GFS except for the plotting routines as described in Fig 1.a.14.

Fig 1.a.14. Characteristics of FNMOC and NOAA/NESDIS QuikSCAT wind and ambiguity displays. Tropical Cyclone Views: Both the Navy and NOAA/NESDIS offer specialized QuikSCAT views for tropical cyclones and possible Suspect Areas (INVEST) as initialized in real-time by either the Joint Typhoon Warning Center (JTWC) or the National Hurricane Center (NHC). The NOAA/NESDIS QuikSCAT Storm Page (seen in Fig 1.a.15) offers a comparison for each Suspect Area between coincidentally plotted wind (at 25 km) and ambiguity plots and high resolution plots of Normalized Radar Cross-Section (NRCS), 12.5 km data and Ultra High-Resolution(UHR)images for each pass over a particular suspect area (see next section for further explanation). The specialized Navy displays can be found in either the Tropical Cyclone pages of FNMOC or the Navy’s Naval Research Laboratory (NRL) (Hawkins et al.) These two sources provide very convenient overlays of scatterometer winds and ambiguities on top of either coincidental IR or Visual imagery or over special-enhanced microwave imagery. These overlays are designed to help the analyst with both ambiguity selection and rain enhancement effects: two of the most difficult concerns for scatterometer interpretation (Fig 1.a.16).

NOAA/NESDIS http://manati.wwb.noaa.gov/quikscat/FNMOC: (PUBLIC ACCESS)

https://www.fnmoc.navy.mil

Wind Vectors Ambiguity Solutions

FNMOC DISPLAY

NRT-NOGAPS Selection

or GFS Selection

Global View (NOAA/NESDIS)

--Near Real Time – NRT (QuikSCAT1)

--GFS model used to aid ambiguity selection

--Available 2-3 hours after overpass

--Advanced Rain Flag technique (black

Wi d V t

Global View (FNMOC – U.S. Navy)

--Data source from NOAA/NESDIS

(NRT - QuikSCAT1)

--Two Model Ambiguity Choices: GFS

NOAA/NESDIS DISPLAY

Ambiguity plotting convention: 1-4 possible

FNMOC: (Public Access) NOAA/NESDIS site

200

Fig 1.a.15. NOAA/NESDIS TC Storm view provides four types of displays. The first 3 (a) containing winds, ambiguity solutions, and NRCS images centered over the latest JTWC or TPC/NHC warning the fourth image (b) is available on a regular basis from the global view page.

a) b)

NOAA/NESDIS Storm Page http://manati.orbit.nesdis.noaa.gov/cgi-bin/qscat storm.pl

201

Fig 1.a.16. Integrated views of QuikSCAT winds and ambiguities over visual (top two) and microwave (85Ghz - middle two) and 85h (bottom) imagery. The ‘X’ marks the center of the circulation.

Microwave 85 GHz QuikSCAT Winds

Visual imagery QuikSCAT Winds

Microwave 85h GHz QuikSCAT Winds

202

Special Ultra High-Resolution (UHR) Data Set at 2.5 km: As briefly described in the previous IWTC 2002 report, the Normalized Radar Cross-Section (NRCS) images (shown above) have an average pixel size of 2.5km with an effective resolution of 6 km. The raw backscatter data offer an amazing view of surface conditions in the vicinity of tropical cyclones (TCs) as well as other high resolution forecasting features of interest. Using some of the same features to create the NRCS images, the BYU Center for Remote Sensing has developed a method, using the four NRCS views to create an ultra high resolution image containing wind vectors (and color coordinated for easier interpretation). Although considerably more noisier than the lower resolution data, they have been found to present great insight into the details of the tropical cyclone structure and are now regularly produced as part of the above tropical cyclone storm data set. These images often detail the changes in wind speed both within and in-between the outer rain bands (Fig 1.a.17).

Fig 1.a.17. New 2.5 km ‘ultra-high’ resolution available over each tropical cyclone invests area. Presented with wind vectors. Understanding how to use QuikSCAT Data: As seen above, integrating the QuikSCAT data with other remote imagery is a good way to determine where rain may be affecting the retrieval process and also provides a manual method to select the proper ambiguity solution. In many cases one can see how the correct ambiguity solutions line up with the cloud lines or rain bands (see Fig 1.a.4, above). Similarly, in areas where rain is suspect of artificially giving a false high wind speed (typically in areas where the real wind speed is less than 30 to 35 kt), overlaying the scatterometer over a recent satellite image will provide a way to see which areas this is likely to happen and which areas it is not (Fig 1.a.18). The method of integrated reconnaissance can work in two ways since a comparison with a satellite image can help determine the correct wind intensity and position for the scatterometer analysis, having a good scatterometer analysis can help the analyst provide a good reference for selecting positions and intensities on poorly defined infrared or visual imagery (Fig 1.a.19).

203

Fig. 1.a.18. Determining intensity by eliminating possible rain-enhanced winds for TS Blanca (02E).

Figure 1.a.19. The scatterometer position is used to help direct the fix for the 37 GHz low level position. It was then translated to a visual image that was closest to the MI data, and then compared to the on-time image that was to support the next warning. Confidence in the fix position is clearly higher than if the IR image was used by itself.

Examples of combined MI and QuikSCAT interpretation in Tropical Cyclones: The following are examples of how combining the QuikSCAT data with that of other recent satellite-based remote sensing data can provide a clearer and more accurate analysis of the tropical cyclone’s position, outer

37 GHz IR Enhanced

204

wind radii, structure and characteristic during both the genesis phase and the extratropical transition phase of development, and finally provide a ‘minimum’ (at least) maximum intensity for those systems just at or below hurricane intensity. In a special case, the high resolution NRCS image is shown to even reveal the structure of the eye wall replacement cycle. a) Positioning: One of the crucial points of any analysis and perhaps one of the easiest to resolve with the combined data (assuming a surface circulation exists). A practical characteristic of many developing depressions is that the low pressure (and circulation) center tends to form within a light wind/rain-free region: two features that are relatively easy to identify in the NRCS and 37 GHz imagery (Fig. 1.a.20).

Fig. 1.a.20. Initial analysis of TD 03E (Carlos) using combined MI (85 and 37GHz) and QuikSCAT data (NRCS, winds, ambiguities). b) Determining a ‘minimum’ maximum TC intensity: This is perhaps one of the most difficult assessments to make. For intense TCs (above ~45-65kt), the analyst must understand the QuikSCAT limiting factors to determine high wind speeds such as the wind retrieval process itself, the 25km resolution of the data, and the tendency for heavy rain regions to attenuate the signal. For weaker TCs (below ~30-40kt), QuikSCAT wind retrievals tend to misinterpret heavy rain as a (incorrect) higher wind speed with a isotopic directional signal that is cross-track to the sensor’s view angle. In this situation, the MI has shown promise in distinguishing the proper wind speeds and directions (see Fig.1.a.6, above). c) Intense Tropical Cyclones and Eye Wall replacement cycle: In this case there is rarely any confusion between the tropical cyclone’s circulation and the environment because the heavy rain and winds coincide. The small ‘dark’ (low rain and low wind region) eye provides for very precise positioning of the TC . Here, the heavy rain pattern around the TC center looks similar to a microwave but often with a smaller central feature. In some cases other characteristics of the TC structure are seen such as concentric eyes or an asymmetric wind field (Fig 1.a.21).

205

Fig. 1.a.21. NRCS views of (a) Super Typhoon Mitag and (b) Typhoon Sinlaku. In each case concentric eyes are evident. Center positions are precisely indicated due to the small calm and rain-free area. d) Determining outer wind radii. Information gained from the analyses in a) and b), above, provide a good starting point for determining outer wind radii. This process is also dependent upon the TC intensity (for determining 50kt (or possibly 64kt) wind radii) and relative location and intensity of the rain bands. In Fig. 1.a.15, an example is shown where the MI may help in these higher wind radii determinations even when there is no overlapping scatterometer data. In weaker TCs a combination of MI and NRCS imagery may help distinguish the artificially high wind speeds in the heavy rain from the expected environmental wind buildup in towards the center as also revealed in Fig. 1.a.22.

Fig. 1.a.22. Wind radii determination in intense TCs.

206

e) Evidence of extratropical transition (ET). Loss of deep (ice seen in the 85 GHz) convection near the TC center and expansion of the winds away from the center and into a horseshoe-like appearance are some of the characteristics of ET in the MI data (Fig. 1.a.23 and 1.a.24).

Fig. 1.a.23. MI characteristics of ET for TY Chan-hom.

Fig 1.a.24. Wind characteristics during the extratropical transition of Hurricane Cindy.

207

e. Future Sensors Microwave imagers will be augmented with the addition of four (4) SSMIS sensors on DMSP spacecraft as noted in Table 1.a.1. These launches will likely occur between 2007-2012, but dates are unknown due to many variables including existing sensor health. There are no microwave imagers on the current NOAA polar orbiter series, nor the European equivalent (METOP-1, 2, 3), but were originally planned for the United States’ converged National Polar-orbiting Operational Environmental Satellite Series (NPOESS). The Conically scanning Microwave Imager Sounder (CMIS) has now been removed from the NPOESS sensor suite and the contract will be recompeted, with launch no sooner than the second satellite (~2016). A total of three (3) microwave imagers may eventually fly on NPOESS and contribute to the global TC structure mapping effort. The Chinese will add the Microwave Radiometer Imager (MWRI) onboard their FY3 spacecraft in the near future. The ten-channel imager will contain 37 and 89 GHz channels that will continue compliment existing capabilities. The French/Indian MADRAS microwave imager is scheduled to fly on the Megha-Tropiques satellite in the 2008-2009 timeframe in a 20-degree tropical inclination. The sensor should provide a wealth of TC views and includes the 37 and 89 GHz channels. In addition, the Japanese will likely fly a GCOM microwave imager similar to AMSR-E in ~ 2012 that would provide excellent AMSR-E like resolution microwave channels. NASA’s Global Precipitation Mission (GPM) includes a microwave imager (GMI) and a dual polarization radar (DPR) on the same platform, similar in context to the current TRMM sensor suite (Smith, et al, 2004 and Hou et al., 2005). The combined imager and radar in a 66 degree inclination orbit will provide the superb spatial resolution required to capture TC structure and sorely needed cross calibration with other microwave imagers. In addition, a second spacecraft will contain a GMI in a tropical inclination that will enhance the opportunities to fly over TCs on multiple consecutive orbits. GPM launch is tentatively slated for the 2013 timeframe. EUMETSAT is studying the need for a microwave imager on future METOP spacecraft and might add one for METOP-4. If so, this would provide a welcome addition to the constellation in the mid-morning orbit. Microwave sounders have been routinely utilized as “imagers” by creating images from their 89 GHz channels. Although, the spatial resolution is poor and varies across the swath, the huge 2350-km swath and three operational sensors provide huge potential advantages. While the spatial resolution is inadequate to resolve many TC features, the storm “context” they provide and decent resolutions at nadir often provide valuable input. Data from the Advanced Microwave Sounding Unit (AMSU-B) and the new Microwave Humidity Sounder (MHS on NOAA-18) have proven beneficial. This bodes well for continued use since both NPOESS and METOP carry microwave sounders with 89 GHz channels. In addition, other countries are slated to contribute additional microwave sounders that should assist in temporal sampling enhancements.

208

Table 1.a.1: A listing of past, present and future passive microwave imagers potentially available for tropical cyclone monitoring. Note that launch dates are highly variable and subject to frequent changes and delays (NRC, 2006). f. Training efforts An excellent introduction for forecasters and newcomers is the Cooperative Program for Operational Meteorology, Education, and Training (COMET) module available online: http://meted.ucar.edu/npoess/tc_analysis/. Taking a global approach, it introduces passive microwave imaging, discusses elementary techniques for fixing storms, explains the difference between passive and active remote sensing, covers conical vs. cross-track sensing techniques, introduces several special cases (double eye wall, cirrus covered eye, low-level detached center, etc.), and surveys the changing state of polar satellites and sensors. A main purpose is to train forecasters at facilities like the National Hurricane Center and the Joint Typhoon Warning Center. Figure 1.a.25 contains an example graphic, highlighting a rare tropical cyclone moving into Brazil. In addition, the Naval Research Laboratory maintains several tutorials at: http://www.nrlmry.navy.mil/training-bin/training.cgi. (Choose “Tropical Cyclones” on the left frame and then select between SSM/I and TRMM.) These tutorials contain detailed case studies to illustrate passive microwave image interpretation. The topics include polarization correction; images of 85 vs. 37 GHz; use of 4-panel composites; interpretation of rain, water vapor, and wind speed products; and the distinction between ice and liquid precipitation within storms. Special attention is given to the detection of storm centers in microwave products.

209

Figure 1.a.25: COMET/NRL web-based graphic illustrating the use of TMI 85 GHz polarization corrected temperature (PCT) to extract storm structure while system is landfalling into Brazil on 28 March 2004 at 0501 GMT. 1.a.5. Summary and Recommendations The length of this report is an indication of the advances that have been made in satellite remote sensing of TC structure since the last IWTC. We are better understanding the physics and signals from the MW/scatt data, and IR data analyses are trending more towards objective techniques. The next four years will bring a focus on multi-spectral data integration to take advantage of each sensor’s capabilities. It should be an exciting time for researchers and forecasters as we tackle TC intensity and structure monitoring from space.

REFERENCES: Atlas, R., R.N. Hoffman, S.M.Leidner, J. Sienkiewicz, T.-W. Yu, S. C. Bloom, E. Brin, J. Ardizzone, J.Terry, D. Bungato, and J. C. Jusem, 2001: The effects of marine winds from scatterometer data on weather analysis and forecasting. Bull. Amer. Meteor. Soc.,82, 1965-1990. Beven, J. and C. Velden, 2003: Satellite Signatures of Extratropical Transition. CD of Workshop Presentations, 2nd International Workshop on Extratropical Transition, Halifax, NS.

Brueske, K.B., and C. S. Velden, 2003: Satellite-Based Tropical Cyclone Intensity Estimation Using the NOAA-KLM Series Advanced Microwave Sounding Unit (AMSU). Mon. Wea. Rev., 131, 687-697.

Cecil, D.J. and E.J. Zipser , 1999: Relationships between Tropical Cyclone Intensity and Satellite-Based Indicators of Inner Core Convection: 85-GHz Ice- Scattering Signature and Lightning, Mon. Wea. Rev., 127, 103–123. Cram, T. A., J. A. Knaff, M. DeMaria, and J. P. Kossin, 2006: Objective identification of annular hurricanes using GOES and reanalysis data. 27th Conf. on Hurricanes and Tropical Meteorology, Monterey, CA, 24-28 April 2006, Amer. Meteor. Soc.

210

DeMaria, M., M. Mainelli, L.K. Shay, J. A. Knaff, J. Kaplan, 2005: Further Improvement to the Statistical Hurricane Intensity Prediction Scheme (SHIPS). Wea. and Forecasting, 20:4, 531-543. Demuth, J. L., M. DeMaria, J. A. Knaff, T. H. Vonder Haar, 2004: Evaluation of Advanced Microwave Sounding Unit Tropical-Cyclone Intensity and Size Estimation Algorithms. J. Appl. Meteor., 43, 282-296. Demuth, J. L., M. DeMaria, J. A. Knaff, T. 2006: Improvements of Advanced Microwave Sounding Unit Tropical-Cyclone Intensity and Size Estimation Algorithms. Submitted to J. Appl. Meteor. Dvorak, V.F., 1975: Tropical cyclone intensity analysis and forecasting from satellite imagery. Mon. Wea. Rev., 103, 420-430. Dvorak, V.F., 1984: Tropical cyclone intensity analysis using satellite data. NOAA Technical Report NESDIS 11, NTIS, Springfield, VA 22161, 45pp. Edson, R.T., M.A. Lander, C.E. Cantrell, J.L. Franklin, J.D. Hawkins, and P.S. Chang, 2002: Operational use of QuikSCAT over tropical cyclones. The 25th Conference on Hurricanes and Tropical Meteorology. San Diego, CA, Amer. Meteor. Soc.,41–42. Edson, R.T., 2002a. Use of scatterometer data in tropical cyclone forecasting and research. Imminent uses and data assimilation, Topic 0.2. Proceedings of the Fifth WMO International Workshop on Tropical Cyclones (IWTC-V), Cairns, Queensland, Australia. 3-12 December 2002. WMO/TD No. XXX. Edson, R.T., 2002b. Tutorial on QUIKSCAT. Special Focus Topic 1.b. Proceedings of the Fifth WMO International Workshop on Tropical Cyclones (IWTC-V), Cairns, Queensland, Australia. 3-12 December 2002. WMO/TD No. XXX. (in preparation). Edson, R.T. and P.S. Chang, 2003 : Normalized radar cross-section patterns from QuikSCAT—A new analysis tool over the tropical ocean. Proc of the 12th Conf on Satellite Meteor. and Oceanography, Long Beach, CA. Edson, R.T. and M.A. Lander, 2003: A method for integrated satellite reconnaissance fix accuracy. Proceedings of the 12th Conference on Satellite Meteorology and Oceanography, Long Beach, CA. Gaiser, P. G., K. M. St. Germain, E. M. Twarog, G. A. Poe, W. Purdy, D. Richardson, W. Grossman, W. L. Jones, D. Spencer, G. Golba, J. Cleveland, L. Choy, R. M. Bevilacqua, and P. S. Chang, 2004: The WindSat spaceborne polarimetric radiometer: Sensor description and early orbit performance, IEEE Trans. Geosci. Remote Sensing, 42, 11, 2347-2361. Goldberg, M. D., D. S. Crosby, and L. Zhou, 2001: The limb adjustment of AMSU-A observations: Methodology and validation. J. Appl. Meteor., 40, 70-83. Halterman, R. R., and D.G. Long, “A Comparison of Hurricane Eye Determine Using Standard and Ultra-High Resolution QuikSCAT Winds,” Proceedings of the International Geoscience and Remote Sensing Symposium, Denver, Colorado, 31 Jul. – 4 Aug. 2006. Hawkins, J. D., T. F. Lee, K. Richardson, C. Sampson, F. J. Turk, and J. E. Kent, 2001, Satellite multi-sensor tropical cyclone structure monitoring, Bull. Amer. Meteor. Soc., Vol. 82, 4, 567-578. Hawkins, J. D., T. F. Lee, F. J. Turk, K. L. Richardson, S. Miller, C. Sampson, J. Kent, and R. Wade, 2006, NRL tropical cyclone R&D web resource augmentations, 14th AMS Conference on Satellite Meteorology and Oceanography, CD-ROM.

211

Herndon, D.C., and C. S. Velden, 2004: Upgrades to the UW CIMSS AMSU-based Tropical Cyclone Intensity Algorithm, 26th AMS Conf. On Hurricanes and Tropical Meteorology, CD-ROM. Hoshino, S. and T. Nakazawa, 2006: Estimation of Tropical Cyclone’s Intensity Using TRMM/TMI Brightness Temperature Data, J. Meteor. Soc. Japan. (submitted) Hou, A. Y., 2005, Update on the Global Precipitation Measurement (GPM) Mission, Presented to the National Academies Committee on the Future of Rainfall Measuring Missions, Washington, D.C., October 18, 2005. Jiang, H., and E. J. Zipser, 2006, Retrieval of hydrometeor profiles in tropical cyclones and convection from combined radar and radiometer observations, J. Applied Meteor., In press. Jones, W.L., I. Adams, J. D. Park, and S.S.Chen, 2002: Evaluation of seawinds wind speed measurements. The 25th Conference on Hurricanes and Tropical Meteorology. San Diego, CA, Amer. Meteor. Soc.,555–556. Kidder, Q, S., W. M. Gray, T. H. Vonder Haar, 1978: Estimating tropical cyclone central pressure and outer winds from satellite microwave Data. Mon. Wea. Rev.., 106, 1458-1464. Knaff, J.A., J.P. Kossin, and M. DeMaria, 2003: Annular Hurricanes. Wea. Forecasting, 18:2, 204-223. Kossin, J. P., 2002: Daily hurricane variability inferred from GOES infrared imagery. Mon. Wea. Rev., 130, 2260-2270. Kossin, J. P., J. A. Knaff, H. I. Berger, D. C. Herndon, T. A. Cram, C. S. Velden, R. J. Murnane, and J. D. Hawkins, 2006a: Estimating hurricane wind structure in the absence of aircraft reconnaissance. Wea. Forecasting, in review. Kossin, J., H. Berger, J. D. Hawkins, and T. Cram, 2006, Development of a Secondary Eyewall Formation Index for Improvement of Tropical Cyclone Intensity Forecasting, Preprints, 60th Interdepartmental Hurricane Conference, Mobile, AL. Lee, T. F., F. J. Turk, J. D. Hawkins, and K. A. Richardson, 2002, Interpretation of TRMM TMI images of tropical cyclones, Earth Interactions E-Journal, 6, 3. Liu, K.S. and J.C.L Chan, 2002: Synoptic flow patterns associated with small and large tropical cyclones over the western North Pacific. Monthly Weather Review, 130, 2134–2142. Lonfat, M., F. D. Marks, and S. S. Chen, 2004, Precipitation distribution in tropical cyclones using the Tropical Rainfall Measuring Mission (TRMM) microwave imager: A global perspective, Mon. Wea. Rev., 132, 1645-1660. Long, D., 2000: Point-wise Wind Retrieval Wind Scatterometry Background. A QuikSCAT/Sigma-0 Browse Product, Ver 2.0. Earth Remote Sensing (MERS), BYU University, Provo, UT., 16 pp. Magnan, S.G., L. E. Carr III, R. L. Elsberry, and M. A. Booth, 1999: Calculating tropical cyclone critical wind radii and storm size using NSCAT winds. Preprints, 23rd Conference on Hurricanes and Tropical Meteorology. Dallas, TX, Amer. Meteor. Soc., 171-174. Mueller, K. J., M. DeMaria, J. A. Knaff, J. P. Kossin, and T. H. VonderHaar, 2006: Objective estimation of tropical cyclone wind structure from infrared satellite data. Wea. Forecasting, in press.

212

Olander, T.L. and C.S. Velden, 2004: The Advanced Objective Dvorak Technique (AODT)-Continuing the journey. Extended Abstracts, 26th Conf. On Hurricanes and Trop. Meteorology, Miami, FL, Amer. Meteor. Soc Olander, T. and C.S. Velden, 2006: The Advanced Dvorak Technique (ADT) -- Continued development of an objective scheme to estimate TC intensity using geostationary IR satellite imagery. Submitted to Weather and Forecasting. Ritchie, E., J. Simpson, W. T. Liu, C. Velden, K. Brueske, and J. Halvorsen, 2002: A closer look at hurricane formation and intensification using new technology. Coping with Hurricanes, R. Simpson, M. Garstang, and R. Anthes, eds., American Geophysical Society, Washington, DC, Chapter 12. Saunders, R. W., T. J. Hewison, and S. J. Stringer, 1995, The radiometric characterization of AMSU-B, IEEE Transactions on Microwave Theory and Techniques, 43 (4), 760. Sharp, R. J., M. A. Bourassa, and J. J. O’Brien, 2002, Early detection of tropical cyclones using SeaWinds-derived vorticity, Bull. Amer. Meteor. Soc., 879-889. Smith, et al., 2004, International Global Precipitation Measurement (GPM) Program and Mission: an overview. In Measuring precipitation from space: EURAINSAT and the future, V. Levizzani and F. J. Turk, eds. Dordrecht, the Netherlands: Kluwer Publishers. Stiles, B.W. and S. Yueh, 2001: Impact of rain on QuikSCAT. Proc. of the AGU 2001 Fall Meeting, 10-14 Dec, San Francisco, CA (also submitted to IEEE Trans. on Geoscience and Remote Sensing). Tenerelli, J.E., S.S. Chen, R. Foster, M. Lonfat, W.T. Liu, and R. Rogers, 2000: Surface winds in Hurricane Floyd: A comparison numerical simulations, aircraft, and QuikSCAT satellite data. Preprints, 24th Conference on Hurricanes and Tropical Meteorology, Ft Lauderdale, FL., Amer. Meteor. Soc., 418-419. Turk, F. J. and S. D. Miller, 2005, Toward improving estimates of remotely-sensed precipitation with MODIS/AMSR-E blended data techniques, IEEE Trans. Geosci and Remote Sensing, 43, 1059-1069. Turk, F. J. and P. Bauer, 2005, The International Precipitation Working Group and its role in the improvement of quantitative precipitation measurements, Bull. Amer. Meteor. Soc., (87), In Press. Velden, C. S., 1989: Observational analysis of North Atlantic tropical cyclones from NOAA Polar-orbiting satellite microwave data. J. Appl. Meteor., 28, 59-70. Velden, C. S., T. Olander, and R. M. Zehr, 1998: Development of an objective scheme to estimate tropical cyclone intensity from digital geostationary satellite imagery. Wea. and Fore., 13, 1720186. Velden, C., et al 2006: The Dvorak Tropical Cyclone Intensity Estimation Technique: A Satellite-Based Method that has Endured for over 30 Years. Bull. Amer. Meteor. Soc., in press. Wacker, R.S., C.S. Velden, and G.W. Petty, 2004: Toward a correction for precipitation scattering effects in satellite-based passive microwave tropical cyclone intensity estimation, 26th AMS Conf. on Hurricanes and Tropical Meteorology, CD-ROM. Wacker, R.S. and C.S. Velden, 2004: Correcting for precipitation scattering effects in satellite-based passive microwave tropical cyclone intensity estimates, 13th AMS Conf. on Satellite Meteorology and Oceanography, CD-ROM. Weissman, D.E., M.A.Bourassa, and J. Tongue, 2002: Effects of rain rate and wind magnitude on Sea

213

Winds scatterometer wind speed errors. J. Atmos. Ocean. Technol., 19,738-746. Wessel, J., J. Cornelius, R. W. Farley, A. Fote, J. Haferman, B. Gardiner, Y. Hong, D. B. Kunkee, G. Poe, S .D. Swadley, D. J. Tesmer, B. H. Thomas, E. Uliana, and D. Boucher, 2004, First observations from the DMSP SSMIS, 13th AMS Conference on Satellite Meteorology and Oceanography. Wimmers, A. J. and C. Velden, "Enhancing satellite imagery through 'morphing' transitions", 13th Conference on Satellite Meteorology and Oceanography, American Meteorological Society, Norfolk, VA, 2004.

Zehr, R., 1989: Improving objective satellite estimates of tropical cyclone intensity. Extended abstracts, 18th Conf. On Hurricanes and Tropical Meteorology, San Diego, CA, Amer. Meteor. Soc., J25-J28.

Zehr, R.M., 2000: Tropical cyclone research using large infrared image data sets. 24th Conference on Hurricanes and Tropical Meteorology. 29 May - 2 June, Fort Lauderdale, FL, Amer. Meteor. Soc., 486-487.

Zehr, R.M., 2003: Environmental vertical wind shear with Hurricane Bertha (1996). Wea. Forecasting, 18:2, 345-356.

214

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES Topic 1b : Special Focus on Field Experiments related to Tropical Cyclone structure: The Hurricane Rainbands and Intensity Change Experiment (RAINEX) and Coupled Boundary Layer Air-Sea Transfer (CBLAST)-Hurricane Experiment1 Topic Co-Chair (CBLAST): P. G. Black NOAA/AOML-Hurricane Research Division 4301 Rickenbacker Causeway Miami, FL 33149 E-mail: peter.black@noaa.gov Fax: 305-361-4402 Topic Co-Chair (RAINEX): Shuyi S. Chen

University of Miami Rosenstiel School of Marine and Atmospheric Sciences Meteorology and Physical Oceanography Division 4600 Rickenbacker Causeway Miami, FL 33149

E-mail: schen@rsmas.miami.edu Fax: 305-421-4696 Working Group (CBLAST): P. G. Black, W. M. Drennan, J. R. French, C. W. Fairall, E. J. Walsh, E. A. Terrill, K. Melville Collaborators (CBLAST): E. A. D’Asaro, P. P. Niiler, T. B. Sanford, J. A. Zhang, J. Kleiss, W. Asher, J.-W. Bao, J. Wilczak, M. Banner, R. Morrison Working Group (RAINEX): S. S. Chen, R. A. Houze, Jr., W.-C. Lee, B. Smull, D. Nolan

1.b.1 Introduction Tropical Cyclone structure and intensity change are the central issue for improving storm intensity forecasts. The devastating 2004-2005 Atlantic hurricane seasons and storm-induced flash flooding in the southeastern Asia highlighted the urgent needs for better forecasts of storm structure and intensity. The lack of skill in forecasts of intensity may be attributed in part to deficiencies in the current operational prediction models: 1) insufficient grid resolution to resolve the inner core (eye and eyewall) and outer rainbands, 2) inadequate surface and boundary layer formulations, and 3) the lack of full coupling to the ocean. A key to improve intensity forecasts is a better understanding the internal dynamic processes influencing rapid intensity change such as eyewall replacement cycle. The extreme high winds, intense rainfall, large ocean waves, and copious sea spray in hurricanes push the surface-exchange parameters for temperature, water vapor, and momentum into untested regimes. Two recent field programs are designed to address these specific questions related to structure and intensity change in tropical cyclones.

1 A portion of the material in this report will appear in two forthcoming issues of the Bull. Amer. Meteorol. Soc.

215

The Hurricane Rainband and Intensity Change Experiment (RAINEX) is aimed to address the hurricane internal dynamics via intensive aircraft observations and high-resolution numerical modeling. The Coupled Boundary Layer Air-Sea Transfer (CBLAST)-Hurricane program is aimed at developing improved coupling parameterizations, using the observations collected during the CBLAST-Hurricane field program, for the next generation hurricane research prediction models. Hurricane induced surface waves (that determine the surface stress) are highly asymmetric, which can affect storm structure and intensity significantly. The stress is supported mainly by waves in the wavelength range of 0.1-10 m, which are the unresolved “spectral tail” of present wave models. The CBLAST-Hurricane modeling team developed a wind-wave parameterization that includes effects of the wave spectral tail on the drag coefficient. Part I The Hurricane Rainband and Intensity Change Experiment (RAINEX) 1.b.2 RAINEX The Hurricane Rainband and Intensity Change Experiment (RAINEX) used three P3 aircraft aided by high-resolution numerical modeling and satellite communications to investigate the 2005 Hurricanes Katrina, Ophelia, and Rita (Houze et al. 2006). The aim was to increase the understanding of tropical cyclone intensity change by interactions between a tropical cyclone’s inner core and rainbands. All three aircraft had dual-Doppler radars, with the ELDORA radar on board the Naval Research Laboratory's P3 aircraft, providing particularly detailed Doppler radar data (Lee 2006). Numerical model forecasts helped plan the aircraft missions, and innovative communications and data transfer in real time allowed the flights to be coordinated from a ground-based operations center (Chen 2006). 1.b.2.1 RAINEX science plan A basic premise of RAINEX was that tropical cyclone intensity changes are associated at least in part with the interactions of the mesoscale circulations of eyewalls and rainbands. Therefore, the RAINEX flight program aimed to obtain mesoscale air motions and thermodynamics via dual-Doppler radar and intensive dropsonde data within and in the immediate vicinity of the most prominent rainbands and eyewalls that presented themselves on the lower-fuselage radars of N42 and N43. Figure 1.b.2.1 illustrates the basic RAINEX flight strategy by superimposing basic flight track modules on the idealized eyewall/rainband pattern of Willoughby (1988). The RAINEX science objectives are 1) using airborne observations to examine simultaneously the dynamic and thermodynamic structures of hurricane inner core and outer rainband regions where the positive potential vorticity associated with deep convective cores are located, and 2) using numerical model to investigate the interactions of the rainbands and primary hurricane vortex circulation and their role in hurricane intensity change.

216

Fig. 1.b.2.1 Idealized flight track plans for the P3 aircraft in RAINEX. The idealized tracks are overlaid on the schematic hurricane radar echo pattern of Willoughby (1988). 1.b.2.2 Aircraft facilities RAINEX employed three P3 aircraft, equipped with Doppler radar and dropsonde capability (Fig. 1.b.2.2). All three aircraft were based at MacDill Air Force Base in Tampa, Florida, at the headquarters of the National Oceanographic and Atmospheric Administration (NOAA) Aircraft Operations Center (AOC). Flights were controlled from the RAINEX Operations Center (ROC) at the Rosenstiel School of Marine and Atmospheric Science (RSMAS) of the University of Miami (UM). Two of the aircraft participating in RAINEX aircraft were the NOAA P3 aircraft, referred to as N42 and N43. The third P3 aircraft in RAINEX was the U. S. Naval Research Laboratory (NRL) P3. The dual Doppler radar system on the NRL P3 was the National Center for Atmospheric Research (NCAR) dual-beam ELDORA radar (Hildebrand et al. 1996; Wakimoto et al. 1996), which is noted for its fine horizontal sampling resolution of about 0.4 km. In RAINEX, N43 was equipped with the original NOAA single parabolic antenna, which accomplished dual-Doppler observation by alternately scanning the antenna fore and aft, while N42 carried two French-built flat-plate antennas which scanned at a fixed 20 degrees fore and aft of the plane perpendicular to the fuselage, and accomplished dual-Doppler observations by switching transmission between antennas during successive scans (Frush et al. 1986; Hildebrand 1989; Jorgensen and Smull 1993; Jorgensen et al. 1996). The along-track sampling resolution is about 1.5 km with the NOAA P3 radars. These along-track sampling rates lead to horizontal resolvable wavelengths of 2 and 8 km for the dual-Doppler analyses with the ELDORA and NOAA P3 aircraft, respectively. RAINEX was the first time that the higher-resolution ELDORA radar has been used in tropical cyclones. All three aircraft were supplied with enough dropsondes to obtain soundings at intervals of 5-10 min when the radars were observing rainbands and eyewalls. NRL, NOAA, and NCAR engineering support staff for the aircraft and instruments were located primarily at MacDill but also in Miami and Boulder.

217

Fig. 1.b.2.2 RAINEX flight coordination. 1.b.2.3 Real-time high-resolution model forecasts for mission planning The aircraft measurements were supported by real-time high-resolution numerical model forecasts of each hurricane. These forecasts were performed with the UM high-resolution, vortex-following, coupled ocean-atmosphere version of the fifth generation PSU/NCAR non-hydrostatic mesoscale model (MM5) coupled with a wave model and an ocean model. A mini-ensemble of the MM5 5-day forecasts was made daily using large-scale model forecasts from four different operational centers as initial and lateral boundary conditions. The NCEP Global Forecast System (GFS), the Naval Operational Global Atmospheric Prediction System (NOGAPS), the Canadian Meteorological Center (CMC), and the Geophysical Fluid Dynamic Lab (GFDL) large-scale models were included. In addition, experimental forecasts with the NCAR Weather Research Forecast (WRF) model were conducted at RSMAS/UM and NCAR. MM5 forecasts were made when there was an active tropical storm with the vortex-following nested grids at 15, 5, and 1.67-km resolution. Fig. 1.b.2.3 shows an example of the mini-ensemble model forecasts in Hurricane Katrina in real-time. It shows a remarkable improvement in skill for both the MM5 and WRF model prediction of intensity in Hurricane Katrina when the grid resolution was increased to 1-2 km. The model forecast output was used daily in flight planning for RAINEX. In post-analysis, high-resolution model simulations of the hurricanes observed during RAINEX will be conducted using the fully coupled atmosphere-wave-ocean model developed at RSMAS/UM (Chen et al. 2006). The aircraft data will be compared to the model output to determine the accuracy of the model real-time forecasts and simulations. The model will thus extend the diagnosis beyond what would be possible from the data alone.

218

Fig. 1.b.2.3 Mini-ensemble of MM5 (red solid lines) and WRF (brown solid lines) forecasts of the maximum wind speed in Hurricane Katrina at 15, 5, and 1.67 km resolutions using various large-scale model forecast fields as lateral boundary conditions. The large-scale model forecasts are in dashed lines (GFS-blue, CMC-magenta, NOGAPS-cyan, and GFDL-red). The models are initialized at 0000 UTC 27 August 2005. 1.b.2.4 Hurricane Rita (2005) Here we illustrate the RAINEX data set with Hurricane Rita (Fig. 1.b.2.4), in which data were obtained on five successive days, with flights documenting the Tropical Storm stage, the rapid intensification to Category 5, an eyewall replacement, and the conversion to asymmetric storm structure when the hurricane encountered environmental wind shear. Preliminary analysis of the high-resolution ELDORA data in Rita shows convective elements oblique to rainband and eyewall structures, consistent with a secondary eyewall with small scale internal features sheared into narrow filaments by the radially varying azimuthal wind (Fig. 1.b.2.4). Preliminary analysis of the dropsonde data obtained in the eyewall replacement phase indicated that the moat between the inner and outer eyewalls was developing into a new eye region (Fig. 1.b.2.5).

219

Fig. 1.b.2.4 RAINEX flight tracks in Hurricane Rita (2005), NOAA N43 (red), NOAA N42 (cyan), and NRL P3 (blue).

Fig. 1.b.2.5 ELDORA radar observed vortcity.

220

Fig. 1.b.2.6 (a) ELDORA radar reflectivity on 22 September 2005 and (b) Dropwinsonde in the moat region (as indicated by a white dot in (a)) of Rita showing “eye-like sounding”.

221

1.b.2.5 RAINEX Summary RAINEX collected unprecedented airborne data in 2005 Atlantic Hurricanes Katrina, Ophelia, and Rita. RAINEX was the first experiment in which three airborne dual-Doppler radar systems were used in hurricanes. It was also the first experiment in which the higher-resolution ELDORA radar was used in a hurricane. These radar data were accompanied by upward of 1000 soundings, including 600 dropsondes targeted with the help of the ground-based operations center for optimal coordination with the airborne dual-Doppler radar observations. The comprehensive RAINEX dataset is available via the NCAR Field Catalog and RAINEX Data Archive. A mini-ensemble forecast product set was provided in real time during RAINEX with the UM vortex-following high-resolution (~1.67 km in the inner domain) modeling system. These forecasts were remarkably accurate, reproducing both the rapid intensification of Katrina and a version of the eyewall replacement as well as the vertical wind shear in Rita. The forecasts were particularly useful in flight planning. The general accuracy of the forecasts bodes well for more detailed analysis of model simulations, taking into account the extensive sounding dataset of RAINEX. These improved simulations can then be compared with the airborne Doppler-radar data and used to analyze the mesoscale generation of PV in eyewalls and rainbands, as was the goal of RAINEX. The RAINEX dataset will provide a basis for a wide range of hurricane studies over the next several years. The level of detail in the radar data, the positioning of the aircraft relative to rainbands and eyewalls, and the targeted dropsondes should provide a basis for unravelling the nature of rainbands, their interactions with eyewalls, and the relation of tropical cyclone internal structure to hurricane intensity. The storms were observed in all stages of development, from Tropical Depression to Category 5 hurricane. The data from RAINEX are readily available through an online Field Catalog and RAINEX Data Archive. The RAINEX dataset is illustrated in Houze et al. (2006) by a preliminary analysis of Hurricane Rita, which was documented by multi-aircraft flights on five days: 1) while a tropical storm, 2) while rapidly intensifying to a Category 5 hurricane, 3) during an eyewall replacement, 4) when the hurricane became asymmetric upon encountering environmental shear, and 5) just prior to landfall.

222

Part II The Coupled Boundary Layer Air-Sea Transfer (CBLAST) experiment 1.b.3 CBLAST-Hurricane The Coupled Boundary Layer Air-Sea Transfer (CBLAST) experiment was conducted during 2000-2005 to improve our fundamental understanding of physical processes at the air-sea interface. We focus here on the CBLAST-Hurricane component, which included experimental observations of the air-sea exchange process in high winds suitable for improving hurricane track and intensity model physics. Other CBLAST activities focused on low wind dynamics (Edson et al 2006) and coupled modeling of hurricanes (Chen et al 2006).

Energy exchange at the air-sea interface is one of three major physical processes governing hurricane intensity change. The others are environmental interactions with surrounding large-scale features in the atmosphere and internal dynamics such as eyewall replacement cycles and cloud microphysics. The air-sea exchange of heat, moisture and momentum determines how hurricanes gain their strength and intensity from the ocean. This has become an extremely important problem over the past several years as we have entered a new era of greater numbers of hurricanes (Goldenberg, 2001), as well as an era of more intense hurricanes (Emanuel, 2005; Webster et al., 2005; Landsea, 2005). The past two years have witnessed an increase in the number of major hurricane landfalls. While efforts to forecast hurricane track have improved greatly over the past 15 years, our ability to forecast hurricane intensity has shown little skill (DeMaria et al., 2005). With more hurricane threats on the U. S. and Caribbean coastlines, the effort to improve hurricane intensity forecasting has taken on greater urgency. The mitigation actions that are taken by emergency management officials, local, state and federal governments and private industry all depend on predictions of intensity thresholds at and near landfall. In response to this need for improved hurricane intensity forecasts, the Office of Naval Research (ONR) initiated the CBLAST program to complement ongoing hurricane intensity research programs in universities and government laboratories such as the Hurricane Research Division. The resulting CBLAST Hurricane experiment became a cooperative undertaking between the ONR, NOAA’s Office of Oceanic and Atmospheric Research (OAR), Hurricane Research Division (HRD), Aircraft Operations Center (AOC), including it’s United States Weather Research Program (USWRP) and the United States Air Force Reserve Command’s (AFRC’s) 53rd Weather Reconnaissance Squadron (WRS). ONR provided support for 17 principal investigators from universities and government laboratories. NOAA provided aircraft flight hour support for two WP-3D research aircraft, expendable probes and Hurricane Field Program infrastructure. AFRC, through the 53rd WRS, provided infrastructure support, specialized expertise in air deployment of large platforms and WC-130J and C-130J aircraft support. The observational strategies and initial results of this effort are described in the following pages. The overarching goal of CBLAST was to provide new physical understanding that would improve forecasting of hurricane intensity change with the new suite of operational models now undergoing testing and evaluation at the Naval Research Laboratory (NRL) and at NOAA’s Environmental Modeling Center (EMC). CBLAST focused an intensive effort on observing air-sea interaction processes within hurricanes because of the recognized lack of knowledge of the physics of air-sea exchange at winds above gale force. Prior to CBLAST, no in-situ air-sea flux measurements existed at wind speeds above 22 ms-1. Parameterization schemes used to approximate air-sea transfer at hurricane wind speeds were simply an extrapolation of low wind measurements with the assumption that the physical processes were the same – despite clear evidence to the contrary! A key goal of CBLAST was to extend the range of observations for exchange coefficients of momentum, heat and moisture across the air-sea interface to hurricane force winds, and above.

223

1.b.3.1 CBLAST Concept and Observation Plan The CBLAST experimental design consisted of two major observational components: 1) airborne in-situ and remote sensing instrumentation flown into hurricanes by the two NOAA WP-3D aircraft and 2) air-deployed surface drifting buoys and subsurface profiling floats. This was intended to provide a mix of ‘snapshots’ of inner-core hurricane conditions each day over a 2-4 day period together with a continuous time series of events at particular ocean locations. A third component, available based on operational needs, consisted of the hurricane synoptic surveillance program designed for improved track forecasting. It provided, on occasion, concurrent high-level NOAA G-IV jet aircraft flights in the hurricane environment, deploying GPS dropsondes to profile the steering currents and significant synoptic features, in addition to reconnaissance flights within the hurricane’s inner core from the WC-130H aircraft, operated by AFRC 53rd WRS. Polar-orbiting and geostationary satellite platforms provided additional remote sensing measurements in the hurricane’s inner core and environment. This approach provided an overarching data base to allow intensity changes from air-sea interaction causes to be separable from those due to atmospheric environmental interactions and internal dynamics. The aircraft component of CBLAST had two modules: a) an aircraft stepped descent module and b) an inner-core survey module. The former was designed to focus on in-situ air-sea flux and spray measurements, while the latter was to focus on large-scale structure, eyewall flux budget measurements and documentation of internal dynamics. The centerpiece of this effort involved a multi-sonde sequence of 8-12 GPS dropsondes dropped from coordinated WP-3Ds flying in tandem at different altitudes across the hurricane eyewall and eye (Fig. 1.b.1). Each module consisted of several options related to precise experimental patterns dictated by prevailing conditions and available time on station. For instance, the stepped descents (Fig. 1.b.2), designed to probe the hurricane boundary layer down to as low as 70m above the sea, were only carried out in clear air conditions between rainbands. Both modules were complemented with an array of airborne remote and in-situ sensors. Air-deployed drifting buoys (drifters) and oceanographic floats (auto-profiling oceanographic radiosondes) were designed to further complement the airborne in-situ and remote sensing of the air-sea interface. This drifter/float air-deployment module consisted of arrays of sensors measuring continuous time series of surface and upper ocean conditions before, during and after hurricane passage. Together the aircraft and drifter/float array provided a unique description of air-sea fluxes, surface wave and upper ocean conditions in hurricane conditions never before achieved. In a similar fashion, the inner core survey module provided observations of significant changes in the inner core dynamics occurring concurrently with observed air-sea processes.

224

Figure 1.b.1. CBLAST survey pattern showing planned expendable probe deployments along a ‘figure 4’ pattern relative to the storm’s eyewall and rainband features. Location of planned stepped-descent patterns to measure boundary layer fluxes is shown schematically. IP is the initial point in the pattern, and FP is the final point.

225

Figure 1.b.2. Vertical alignment of stepped descent flight legs along with expendable probe location

along the 25 nmi (46 km) leg length.

1.b.3.2 Case Studies The CBLAST experimental effort began in 2000 with the development of seven new airborne instrument systems, three new oceanographic float designs, 2 drifting buoy designs, the flight pattern strategy and the air-deployment strategy, including the WC-130J air-deployment certification and air-drop certification of 3 platform types. The new airborne instrument systems were 1) Best Aircraft Turbulence (BAT) probe for fast response temperature and u-,v-,w-wind components, 2) a modified LICOR fast response hygrometer, 3) CIP particle spectrometer and 4) Particle Doppler Analyzer (PDA) for sea spray droplet observation, 5) Scripps downward-looking, high-speed visible and infrared video camera systems for wave breaking observations, 6) Stepped and Simultaneous Frequency Microwave Radiometers (SFMR and USFMR) for surface wind speed and 7) the Integrated Wind and Rain Atmospheric Profiler (IWRAP) for continuous boundary layer and surface wind vector profiles. These systems were built in 2001 and flight- tested in 2002. Also deployed were three existing systems: 1) Tail (TA) Doppler radar for boundary layer wind structure, 2) Lower Fuselage (LF) weather radar for hurricane precipitation structure and 3) Scanning Radar Altimeter (SRA) for directional wave spectra. Two storms, Edouard and Isidore, were flown in 2002, the first to test the new stepped-descent flight pattern strategy and the second to test extended low level flight pattern for detection of linear coherent turbulence structures, sometimes referred to as ‘roll vortex’ or ‘secondary boundary layer’ circulations. The CBLAST field program began in earnest in 2003 with the survey flight pattern flown on 6 days by the two NOAA WP-3D aircraft (a total of 12 flights, including 12 stepped-descent patterns) in Hurricanes Fabian and Isabel from a staging base in St. Croix, U.S. Virgin Islands. An additional 10 AFRC WC-130H reconnaissance flights and 3 NOAA G-IV surveillance flights were also flown during this period. An array of 16 drifting buoys and 6 floats were deployed by the 53rd WRS from a WC-130J aircraft ahead of Hurricane Fabian. An engine failure due to salt build-up occurred near the end of the sixth flight which resulted in new

226

safety regulations requiring a chemical engine wash after each flight below 340 m. CBLAST flights in 2004 continued, but were restricted to flight levels above the boundary layer. The CBLAST flights in 2004 were in Hurricanes Frances on 4 days, Ivan on 5 days and Jeanne on 3 days. The key success in 2004 was the air-deployment by the 53rd WRS of 38 drifting buoys (30 Minimet; 8 ADOS) and 14 floats (9 ARGO/SOLO; 2 Lagrangian; 3 EM/APEX) ahead of Hurricane Frances on Aug 31. All drifters and floats were deployed successfully. The floats were recovered by the UNOLS ship R/V Cape Hatteras between Sept 27 and Oct 4, approximately 4 weeks after deployment. 1.b.3.3 Key Results from the Aircraft Component

1.b.3.3.a. First turbulence measurements in tropical storm and hurricane force winds The principal results from the aircraft component of CBLAST were the estimation of surface momentum and enthalpy flux from direct eddy correlation measurements using two newly modified airborne instrument packages: the BAT Probe and the LICOR fast response hygrometer. The results are based on measurements obtained during 15 stepped-descent patterns flown in Hurricanes Fabian and Isabel in 2003 (Fig. 1.b.3).

Figure 1.b.3. CBLAST stepped descent flight patterns flown in Hurricanes Fabian and Isabel in 2003, plotted in storm-relative coordinates, with the storm motion indicated by the arrow (up). Circles are shown at 100 km intervals. Flight tracks are superimposed on NASA MODIS visual image of Hurricane Isabel on 14 Sept, 2003 at 1445 UTC. In addition a WP-3D Lower Fuselage (LF) airborne radar image from NOAA 43 of Isabel at 1642 UTC is overlaid indicating typical eyewall and rainband structure. MODIS image courtesy of MODIS Rapid Response Project at NASA/GSFC.

227

These are the first direct flux measurements ever in a hurricane and the first in a tropical storm since the ‘gust probe’ measurements of Moss and Merceret (1976, 1977) and Moss (1978) in the periphery of Tropical Storm Eloise on Sept 17, 19752, where their estimated surface winds were ~ 20 ms-1. Data from a total of 48 (42) flux runs were used to compute independent estimates of the momentum (humidity) flux. All of the flux runs were flown between 70 and 400m at an airspeed of close to 113 ms-1. Measurements from GPS dropsondes indicate that the top of the hurricane planetary boundary layer (PBL), defined as the top of the constant potential temperature layer, was approximately 500 m (Drennan, et al., 2006). Leg lengths ranged from 13 to 55 km, with an average length of 28 km. Fluxes of momentum and humidity were computed using eddy correlation. Profiles of fluxes were analyzed for individual stepped descents. There was no significant height dependence of humidity flux within the PBL. Momentum flux decreased to only 50 to 75 percent of the surface values near the top of the PBL. Surface friction velocity was computed following Donelan (1990). Ten-meter neutral wind (U10N) was taken from leg-averaged measurements (approximately 5 minutes) from a nadir-pointing SFMR (Uhlhorn and Black, 2003; Uhlhorn, et al., 2006). The drag coefficient (CD10N, referred to hereafter as CD) was computed directly from the friction velocity and U10N. Surface saturated specific humidity, q0, was estimated based on the sea surface temperature (SST), provided by an infrared radiometer on the aircraft, averaged along the flight track, and corrected for intervening atmospheric absorption. A linear regression line was fit to the measured SST values at each stair-step altitude and the surface value extrapolated. The difference between the extrapolated surface value and the average value for each flux run was then computed and used to correct the measured SST to the true value. Additional details on this method can be found in Drennan et al. 2006. Neutral ten-meter specific humidity, q10N, was estimated using a logarithmic profile to extrapolate the flight level measurements to 10 m. The moisture exchange coefficient, or Dalton number, (CE10N, referred to hereafter as CE) for each flux run was computed directly from the humidity flux, U10N, q0 and q10N. Details and justification of the above discussion can be found in French et al. (2006) and Drennan et al. (2006). Of course, in an environment where individual wave heights can exceed 20m, the meaning of 10m bulk coefficients should be questioned. We consider them useful as reference values, 10m being the lowest level for many atmospheric models. When CBLAST CD and CE results are compared with other studies (Figs. 1.b.4 and 1.b.5), one can see that they represent a 32 and 61 percent increase, respectively, of the wind speed range of prior observations. It is apparent that the CBLAST high-wind CD values represent a systematic departure from prior estimates. Several surprising results emerged from these measurements. Primarily, CBLAST CD measurements become nearly invariant with wind speed above a 23 ms-1 threshold (Fig. 1.b.5). This is a full 10 to 12 ms-1 less than the hurricane-force threshold of 33 ms-1 obtained using GPS dropsonde measurements by Powell et al. (2003) and laboratory tank measurements by Donelan et al. (2004).

2 While the authors refer to their measurements in ‘Hurricane’ Eloise, the National Hurricane Center Best Track

archives (http://www.nhc.noaa.gov/tracks1851to2005_atl.txt) lists the peak surface winds for Eloise on Sept 17,

1975, when the center was over Hispañola (Hebert 1976), as tropical storm strength, i.e. 22-25 ms-1, for the

18-00 UTC period of the research flight.

228

Figure 1.b.4. Drag coefficient estimates derived from CBLAST stepped-descent flight legs in Hurricanes Fabian and Isabel (2003). The asterisks represent average values in 2.5 ms-1 bins, and the bars show 95% confidence intervals. The squares are from flight legs in the right-front quadrant of the storms, the plus signs from the right-rear quadrant and the diamonds from the left-front quadrant. The black dash-dot line represents the values from Donelan et al. (2004); the red dash-dot line from Powell et al. (2003); blue-dashed line from an average of Smith (1980) and Large and Pond (1981); the red-dashed line from Yelland et al. (1998); black dashed line from HEXOS (Smith et al. 1992), and the grey circles from CBLAST-Low (Edson et al. 2006). CBLAST CD measurements (Fig. 1.b.4) agree with previous open-ocean, gale-force wind measurements in the range 17 - 22 ms-1 for fully developed seas in the North Pacific at Ocean Weather Ship Papa, (Large and Pond, 1981) in the North Atlantic at Sable Island (Smith, 1980) and in the Southern Oceans from a research vessel (Yelland, et al., 1998). Values are lower than CD observations in fetch-limited wave conditions during HEXOS in the North Sea (Smith, et al., 1992), for COARE 3.0 conditions (Fariall et al., 2003) and for CBLAST-low (Edson et al., 2006) . Estimates of CD show little dependence on quadrant of the storm in which the measurements were obtained. However, it should be noted that the natural variability in the data provide less than overwhelming confidence of this result. The value of CD differs little between regions of young, growing waves in the right-rear quadrant in a swell-following environment and regions in the right-front and left-front quadrants where the local sea was older, less steep and higher, with swell at increasing crossing angles with respect to the wind of up to 90 degrees (Wright et al., 2001), a result discussed further in the following sub-section. Another new finding suggests that moisture flux measurements are relatively constant with height within the hurricane boundary layer. Finally, we find that estimates of CE above 20 ms-1 are in good agreement with the results from HEXOS (DeCosmo et al. 1996; modified as per Fairall et al. 2003) and COARE 3.0 (Fairall et al. 2003) extrapolated from 19 ms-1 through our range of measurements (Fig. 1.b.5). This suggests that CE is constant with wind speed to hurricane force winds of 33 ms-1.

229

Figure 1.b.5. Moisture exchange coefficient (Dalton Number) estimates derived from CBLAST stepped-descent flight legs in Hurricanes Fabian and Isabel (2003). The asterisks represent average values in 2.5 ms-1 bins, and the bars show 95% confidence limits. The squares are from flight legs in the right-front quadrant of the storms, plus signs from the right-rear quadrant and diamonds from the left-front quadrant. The black dashed line represents the HEXOS line (DeCosmo et al. 1996), modified as per Fairall et al. (2003) and extended to 36 ms-1. The green solid line is the COARE 3.0 curve (Fairall et al. 2003), and the grey circles are from CBLAST-Low (Edson et al. 2006). It is obvious that while CBLAST extended the wind speed range of prior CD and CE observations, a further increase of the wind speed range is required to validate flux estimates in hurricane-force wind conditions, where physical processes may depart significantly from tropical storm wind conditions as the importance of sea spray and other poorly understood phenomena such as ‘roll-vortex’ features, may increase dramatically. Budget and sea spray studies are underway to estimate CD and CE for hurricane-force conditions by Emanuel (personal communication) using the CBLAST rapid deployment eyewall dropsondes in Fabian and Isabel and by Fairall (personal communication) using laboratory measurements. Emanuel (1986, 1995) has shown that the ratio of CE/CD is an important parameter in estimating hurricane potential intensity. The new CD and CE observations along with the new highly-reliable SFMR surface wind measurements (Uhlhorn and Black, 2003; Uhlhorn, et al., 2006) show CE/CD values to average near 0.7 for tropical storm conditions, slightly below the Emanuel (1995) threshold for hurricane development of 0.75 (Fig. 1.b.6). Relatively new estimates of this ratio from laboratory tank measurements (Fairall, et al., 2006) for winds up to 60 ms-1 suggest that this ratio increases above 45- 50 ms-1 due to spray effects, remaining fairly level near 0.5 for winds of 20 – 45 ms-1, once an upward adjustment of about 0.1 is applied to match COARE 3.0 values. This result also suggests that spray effects may be important mainly in the hurricane eyewall (see section 1.1.4.d for additional discussion).

230

Figure 1.b.6. Ratio of CE/CD derived from CBLAST measurements. The asterisks represent average values in 2.5 ms-1 bins, and the bars show 95% confidence limits. The black dashed curve is the mean ratio from HEXOS (DeCosmo et al. 1996, modified as per Fairall et al. 2003; Smith et al. 1992). The solid green line is the ratio values from COARE 3.0 (Fairall et al. 2003). The grey circles are from CBLAST-Low (Edson et al. 2006). The dark blue, red and light blue lines are for different specifications of the spray droplet source strength when the effects of sea spray are included from the current version of the ESRL sea spray flux parameterization (Fairall and Wilzcak, personal communication). The dash-dot horizontal magenta line is the 0.75 threshold for TC development proposed by Emanuel (1995). CE/CD ratio estimates from budget methodology by Emanuel (personal communication) for axisymmetric mean surface winds of 50 ms-1 in Isabel also suggest that the ratio may increase significantly above CBLAST hurricane values for intense hurricane (≥ CAT 3) conditions. On the other hand, Andreas and Emanuel (2001) have suggested that the role of spray may act to simultaneously increase CD and CE, leaving the ratio nearly equal to our values at high winds. This would require another explanation for how intense hurricanes develop and are maintained. The implications of CE/CD ~ 0.7 for intense storms have been investigated by Montgomery et al., 2006, and Bell and Montgomery (2006) which indicate that ‘superintense’ conditions leading to sustained CAT 4 and 5 conditions, such as observed in Isabel, are a result of strong air-sea interaction inward from the hurricane eyewall leading to augmented horizontal entropy transport via enhanced frictional inflow and eyewall mesovortices. They suggested that this mechanism was a key reason why Isabel maintained CAT 5 status for 3 days.

1.b.3.3.b. Surface wave observations The Scanning Radar Altimeter (SRA) on one of the P3 aircraft recorded huge data sets of wave images and 2D wave spectra in all quadrants of CBLAST storms in Fabian in 2003 and throughout CBLAST

231

storms in 2004. A wave image typical of this data set (Fig. 1.b.7) from the front quadrant of Hurricane Fabian near 500 m flight altitude illustrates the predominant 300 m swell. Superimposed on the swell is the local sea with wavelengths of about 80-100 m crossing at a 90º angle. This is typical of conditions depicted in sector III of Fig. 1.b.9, which illustrate 3 sectors of distinctly different 2D wave spectra in Hurricane Bonnie (1998), discussed by Wright et al., 2001. The spectra in sector I tend to be tri-modal with 2 swell peaks plus the local sea. The spectra in sector II tend to transition from tri-modal to bi-modal with the swell following within 30 degrees of the local sea. The spectra in sector III tend to transition from bimodal to unimodal depending on whether the local sea is resolved. The swell tends to propagate at about a 90-degree angle to the local sea in this region.

Figure 1.b.7. Swath of wave elevations from SRA from 200 m flight altitude during Fabian, 2003. Scale of aircraft is shown at 1 km along track, 0.2 km cross track position. A further illustration of the behavior of the swell relative to the local sea as a function of azimuth is shown in Fig 1.b.8, which shows twelve SRA spectra about 80 km from the eye of Hurricane Ivan, 2004, along with bold lines indicating sectors I. – III. With three distinctly different types of wave spectra. In the right-front quadrant (sector II/III boundary) the wave field is unimodal with 350 m wavelength and 11.4 m wave height. Directly to the right of the track the wavelength shortens to about 260 m and the spectrum broadens and becomes bimodal. In the right-rear quadrant the wave height decreases and the spectrum becomes trimodal (sector I.). In the rear quadrant of Ivan, the wave height and length reach minimum values of 5.6 m and 190 m, about half their values in the right forward quadrant. This suggests the waves are young, steep and short in the right-rear quadrant and older, flatter and longer in the right front and left front quadrants. To the left-rear and left-front of the eye, the wind and waves are about at right angles to each other.

232

The expectation was that the exchange coefficients would exhibit a variability that depended on the characteristics of the 2D wave spectrum. This has turned out to not be the case as one can see by comparing Figs 1.b.4 and 1.b.5 with Fig. 1.b.8. What this says, and what emerges as the second major conclusion for CBLAST hurricane measurements, is that surface fluxes and exchange coefficients derived from them (Figs. 1.b.4-1.b.5) appear not to be a function of the variation in the relationship between the long wavelength swell and the shorter wavelength local sea, at least between tropical storm and hurricane force wind radii.

Figure 1.b.8. The center of the figure shows wind speed contours (ms-1) from the HRD H*WIND surface wind analysis- based mainly on SFMR surface wind speed measurements in Hurricane Ivan at 2230 UTC on 14 September 2004 for a 2° box in latitude and longitude centered on the eye. Arrow at the center indicates Ivan’s direction of motion (330º). Lines separate three sectors (I. – III.) defining distinctly different wave spectra. The storm-relative locations of twelve 2D surface wave spectra measured by the SRA are indicated by the black dots. The spectra have nine solid contours linearly spaced between the 10% and 90% levels relative to the peak spectral density. The dashed contour is

233

at the 5% level. The outer solid circle indicates a 200 m wavelength and the inner circle indicates a 300 m wavelength. The dashed circles indicate wavelengths of 150, 250, and 350 m (outer to inner). The thick line at the center of each spectrum points in the downwind direction, with its length proportional to the surface speed. The upper number at the center of each spectrum is the significant wave height and the lower number is the distance from the center of the eye. The average radial distance for the twelve spectral locations is 80 km. The SRA data which produced the spectra were collected between 2030 UTC on 14 September and 0330 UTC on 15 September. 1.b.3.3.c. CBLAST wind-wave coupling parameterization The coupling of the atmosphere through waves to the ocean is best served by a direct calculation of the evolution of the wave field and the concomitant energy and momentum transfer from wind to waves to upper oceanic layers. The CBLAST-Hurricane modeling team has developed a wind-wave coupling parameterization, based on CBLAST observations, that is tested in a high-resolution (with 1.67 km grid spacing), coupled atmosphere-wave-ocean model (Chen et al. 2006; Zhao and Chen 2006). The coupled model components for the atmosphere, surface waves and the upper ocean circulation are MM5, WAVEWATCH III, and 3DPWD, respectively. The effects of surface wave and ocean coupling on storm intensity are investigated in Hurricane Frances (2004) as shown in Fig. 1.b.3.3.c1. The observed minimum sea-level pressure (MSLP) and maximum wind speed (MWS) from the NHC best track data are compared with the uncoupled and coupled model simulations. The uncoupled MM5, which uses a fixed pre-storm SST throughout of the simulations, overestimates MSLP (an integrated measure of the storm intensity) in all three storms. The coupled A-O, which includes the effect of SST cooling, due mostly to vertical mixing and secondarily to upwelling and direct cooling, produces a more realistic MSLP compared to the uncoupled MM5. However, both the uncoupled MM5 and coupled A-O simulations seem to underestimate MWS compared to the best track data in all three cases. The fully coupled A-W-O simulations are closest to the observed values. The discrepancy in MSLP and MWS in the uncoupled MM5 and coupled A-O simulations are evidently due to the lack of wind-wave coupling compared with the fully coupled A-W-O simulation.

Fig. 1.b.3.3.c1 Observed (the NHC best track in black) and simulated MSLP (dashed lines) and maximum wind speed (solid lines) from the fully coupled atmosphere-wave-ocean model (A-W-O, red), coupled atmosphere-ocean model (A-O, green), and uncoupled atmosphere model (blue), for

234

Hurricanes Frances. The coupling to the ocean circulation model improves the storm intensity by including the storm-induced cooling in the upper ocean and SST, whereas the uncoupled atmosphere model with a constant SST over-intensifies the storms. However, without coupling to the surface waves explicitly, both the uncoupled atmospheric model and the coupled atmosphere-ocean model underestimate the surface wind speed, even though the MSLP of especially the A-O coupled model is close to the observed values. The full coupling with the CBLAST wave-wind parameterization clearly improves the model simulated wind-pressure relationship that is a key issue in hurricane intensity forecasting. 1.b.3.3.d. Sea surface white-capping at high winds Breaking waves producing whitecaps play an important role in air-sea interaction, especially in higher winds and hurricanes. Breaking limits the height of surface waves, it is the main mechanism for the transfer of momentum from wind waves to currents, it is an important source of turbulence at the surface for mixing heat, mass (gas) and momentum, and it is a source of aerosols which may play a significant role in the dynamics and thermodynamics of hurricanes. Furthermore, whitecaps have a significant effect on the microwave brightness temperature of the ocean, and microwave radiometry has proven to be a very useful tool for measuring hurricane winds. Thus an improved understanding of the relationship between breaking and the brightness temperature will lead to improved algorithms for wind retrieval. For all these reasons, an improved knowledge of breaking and white-capping in high winds and hurricanes is essential for an overall improvement in hurricane prediction. During CBLAST we used visible imagery to quantify breaking and whitecap coverage. While CBLAST whitecap and wave data are still being analyzed, the following are some preliminary results of the measurements. During the CBLAST research flights, a nadir-looking mega-pixel digital video camera with a 30 degree field of view on a NOAA P3 captured images of the sea surface at 20 Hz. To infer the whitecap coverage, the background lighting is removed, and images contaminated with clouds are discarded. A brightness threshold is chosen manually by the operator as the minimum threshold that does not include non-breaking areas. The fraction of the image brighter than this threshold represents the fraction of the observed sea surface covered with foam. To differentiate active breaking from older foam, a double-threshold is used. A higher threshold is manually determined to include only the brightest active breaking areas (see Fig. 1.b.9).

235

Figure 1.b.9. Example of the double-threshold processing on an image. The lower threshold in yellow

includes all foam and streaks. The higher threshold in red includes only the most dense, active

patches of bubbles. The scale of the image is 150x100m.

The fraction of the image brighter than the threshold is computed over a 5 minute video sequence, covering roughly 6 square kilometers of sea surface, then averaged together to yield the active (high threshold) and total (low threshold) whitecap coverage for the data segment. The 10-meter wind speed is obtained from the Hurricane Research Division wind analysis, adjusted for the hurricane translation, and location and time of image capture. The whitecap coverage computed using these thresholds is shown in Fig. 1.b.10 along with earlier data from the literature, and recent unpublished data from the Gulf of Tehuantepec.

236

Figure 1.b.10. Whitecap coverage versus 10 meter wind speed. Open squares: CBLAST 20030914 active patches only. Filled squares: CBLAST all foam. Vertical lines connect identical data segments. Blue line with dashed lines indicating spread of data points: compilation of data in Wu (1988). Black line: Monahan et al. (1985). Red line: Hanson & Phillips (1999). Red circles: Melville & Matusov (2002). Blue asterisks: data from Gulf of Tehuantepec (unpublished). It is interesting to compare the earlier data compiled by Wu (1988) which, when extrapolated to larger wind speeds, would show a 100% whitecap coverage at approximately 33 m/s, the threshold wind speed for a Category 1 hurricane on the Saffir-Simpson scale. Our CBLAST data show values lower than the historical data, consistent with the fact that the historical data cannot simply be extrapolated from lower wind speeds. The roll-over of the whitecap coverage may be related to the corresponding behavior in the drag coefficient that was discussed above. Indeed, breaking has been invoked as a mechanism for decoupling the airflow from the waves and modifying the drag coefficient. From the spatial distribution of whitecap coverage (not shown), the highest fraction of breaking occurs close to the hurricane eye wall, in the front right quadrant. This is consistent with measurements of steep waves in the same region. In addition to the video camera, we also had a nadir-looking Riegl laser altimeter which gave the distance from the aircraft to the sea surface at 120Hz. The Riegl only had useful return strength when the aircraft was flying at an altitude below 150m, so there is not much data from the CBLAST missions, especially when altitude restrictions were placed on P3 operations in subsequent years. The data has been interpolated over isolated drop outs, but low wave number aircraft motion has not been removed. The data were interpolated to a regular time step, converted from time to space assuming a constant

237

aircraft speed over the 3-minute record, and Fourier-transformed using a Hanning window. Examples of preliminary surface wave number spectra from Isabel are shown in Fig. 1.b.11. The top sample was directly ahead of the eye of Isabel near the 30 m/s wind contour. The bottom sample was in the front right quadrant on the 40 m/s contour, and the middle sample between the other two. The spectra at the lower wind speeds display an approximate -3 slope, consistent with historical data at lower wind speeds. However, the spectrum at the highest wind speed shows a significant departure from this slope. There are several possible reasons for this including the direction of the aircraft track compared to the wind direction.

Figure 1.b.11. Left panels: raw signal return from fixed-point Riegl altimeter at three locations in Isabel on Sept. 14th, 2003. Right panels: corresponding wavenumber spectra. The observed slope is indicated in each panel (-3.34, -3.11, and -2.68), with circles indicating the peak of the wind-wave spectra at k=.0573, .0692, and .0208, respectively. Lower wavenumbers are due to aircraft motion. This summary shows only preliminary results of the analysis of the CBLAST video and altimeter data. A more detailed analysis of the data is underway to investigate the relationship between the whitecap data and surface emissivity from the Stepped Frequency Microwave Radiometer (SFMR) data collected during CBLAST.

238

1.b.3.3.e. Sea spray effects on hurricane surface fluxes

For the last decade the NOAA Environmental Sciences Research Laboratory (ESRL) and collaborators have been developing a hierarchy of models of the production sea spray at high winds and the subsequent thermodynamic effects of the evaporation of spray on hurricane boundary layers. There are three steps in this process:

1) Characterization of the size spectrum of droplets produced by the ocean as a function of the forcing (wind speed, stress, wave breaking, etc),

2) Computation of the exchanges of heat and moisture between the droplets and an unperturbed

near-surface layer structure, and

3) Accounting for the ‘subgrid-scale’ distortion of the standard surface layer T/RH structure by the droplets (a process referred to as ‘feedback’).

Recent work done under the CBLAST umbrella has included:

• Theoretical advances connected with improving the parameterization • Observations of sea spray droplet spectra from the NOAA WP-3 aircraft during the

CBLAST field programs • A laboratory study of the scaling of the droplet surface source spectrum • Investigation of the impact of the parameterization when implemented in a NWP hurricane

model. • Sensitivity studies of droplet source strength coupled to the hurricane boundary layer using

an droplet-explicit 1-D turbulent closure model In 2003 droplet observations were made by ESRL during the CBLAST flights used the Cloud Imaging Probe, CIP from Droplet Measurement Technologies (DMT). The CIP is a technology based on a linear array of light detecting diodes. Droplets are sized from 25 to 1550 µm diameter in 62 equally spaced 25 µm diameter bins. Droplet data were obtained in low-altitude level flight segments (step descents) nominally between rainbands in hurricanes Fabian (Sept. 2, 3, and 4) and Isabel (Sept. 12 and 13). Steps were done at altitudes ranging from 66 to 760 m; wind speeds at 100 m were typically 29 ms-1 (equivalent to a 10-m wind speed of 24 ms-1). In Fig. 1.b.12 we show concentration spectrum of all droplets counted in the 10 usable profiles at altitudes less than 250 m (about 1 hour of data) where the red line is the sum of two exponential distributions with radius modes of 10 and 100 µm. The concentrations at altitudes on the order of 100 m are consistent with near-surface measurements (e.g., De Leeuw, 1990), but the mean curve implies many more large droplets (radius > 500 µm).

239

Figure 1.b.12. Droplet concentration spectrum combining CIP data from10 level runs below 250 m altitude in Hurricanes Fabian and Isabel: o - median; x - mean. The University of Washington Phase Doppler Anemometer (PDA) system was installed on the NOAA P-3 in December 2003 and a test flight showed very good comparison with the ESRL CIP. In 2004 the PDA was flown through both Hurricane Frances and Hurricane Jeanne; Fig. 1.b.13 shows a sample droplet spectrum from Jeanne. The 2003 flights were at moderate wind speeds and were not near enough to the surface to encounter the meat of the sea spray layer. The 2004 data are still being analyzed. In 2003, the laboratory study, the Spray Production and Dynamics Experiment (SPANDEX), was conducted at the Wind Tunnel Facility of the Water Research Laboratory in Manley Vale (NSW) in cooperation with the University of New South Wales, Australia. The goals of SPANDEX were to illuminate physical aspects of spume droplet production and dispersion, verify theoretical simplifications presently used to estimate the source function from ambient droplet concentration measurements, and examine the relationship between the implied source strength and forcing parameters such as wind speed, surface turbulent stress, and wave properties. Droplet spectra were measured at different heights above the mean water level with the CIP and PDA. To illuminate the forcing effects, the wind speed, waves, and salinity were varied (see Table 1.b.1). The profile measurements verified the similarity-based balance of surface source strength and gravitational settling and the observed mass flux was close to the Fairall-Banner model (Fairall, et al., 2006 - see below). Careful processing of the small scale surface wave energy revealed a direct correlation of the mass flux with energy going to wave breaking (Fig. 1.b.14). The analysis of results from SPANDEX are nearing completion and a publication is expected to be submitted soon (Fairall et al. 2006).

240

Table 1.b.1. Summary of forcing conditions for the SPANDEX laboratory study.

Fresh water 24 ppt salinity

Uref

(ms-1)

U*a

(ms-1)

Z0

(mm)

U*a

(ms-1)

Z0

(mm)

14.5 1.35 3.68 1.06 1.45

16.7 1.44 3.34 1.64 4.04

17.8 1.64 3.12 1.78 5.25

Figure 1.b.13. A plot of size segregated spray droplet concentrations (i.e., number of droplets in a given size range per volume of air) in a cloud-free area as a function of aircraft altitude for data taken with the phase-Doppler anemometer in Hurricane Jeanne on September 22, 2004 at altitudes of 708 m, 410 m, and 270 m. Droplet concentrations are essentially equal for the two higher altitudes and then increase with decreasing measurement height.

241

Figure 1.b.14. The spray droplet mass flux as a function of energy dissipated by wave breaking from the SPANDEX laboratory study. Our present source strength parameterization is derived from the Fairall-Banner physical sea spray model (which predicts the size spectrum of sea spray produced by the ocean in terms of wind speed, surface stress, and wave properties). The Fairall-Banner spectrum has been parameterized into a simple mass flux representation in terms of friction velocity. The unperturbed thermodynamic effects are based on integrals of the ratios of thermodynamic and suspension time constants following Andreas. Finally, a diagnostic feedback parameterization has been developed which characterizes the way evaporating droplets of various sizes modifies the stratification of the air near the surface, which in turn reduces further droplet evaporation but enhances sensible heat flux carried by the droplets. The present form of the parameterization has two tuning coefficients: one that scales the intensity of the source strength and the other which affects the partitioning of enthalpy flux between sensible and latent heat. The ratio of enthalpy to momentum transfer coefficients scales with wind speed for different choices of the sources strength, as seen previously in Fig. 1.b.6. This result suggests that when spray effects are added, the CE/CD ratio (Fig. 1.b.6) increases with wind speed above approximately 45 - 50 ms-1. Recently the parameterization was coded in F90 and implemented in the GFDL hurricane model and a version of Weather Research Forecast (WRF) model that runs at ESRL. Preliminary tests on hurricanes Ivan and Isabel showed sensitivity to sea spray but there are interdependencies with the non-droplet (direct) transfer specifications in the models. An example for Hurricane Isabel is shown in Fig. 1.b.15, where the current GFDL model simulation was compared with the version incorporating the sea spray parameterization. The GFDL model was run using a 1/6o (~18 km) grid resolution. The simulations were conducted using the coupled GFDL-POM model with no surface waves. In this case the spray parameterization improves the simulation, but in other cases it did not.

242

Figure 1.b.15. Simulations of maximum wind speed for Hurricane Isabel using the GFDL model: blue line, GFDL version 05; green line, version 05 with Fairall-Banner sea spray flux parameterization; black dashed line, observations. ESRL is planning future work to incorporate an explicit microphysics sea-spray scheme (Kepert et al., 1999) within the WRF modeling framework that will be the near-surface equivalent to the explicit cloud microphysics schemes that are presently available in WRF and other atmospheric models. Results from numerical simulations using this explicit microphysics sea-spray scheme will be analyzed to develop, test and calibrate new bulk parameterization schemes that take into account the effects of sea spray droplets, and yet are simple and sufficiently fast to be used in the operational models.

1.b.3.3.f. Evidence for secondary boundary layer circulations Strong evidence was found for the existence of ‘roll vortex’ secondary boundary layer circulations in hurricanes. Complementing the CBLAST flights in 2002-2004 were a number of RADARSAT and ENVISAT Synthetic Aperture Radar (SAR) passes (see Fig. 1.b.16, top) over the storms which all showed streaks in the radar backscatter with wavelengths on the order of 800-1000 m. These were most prominent in the front semicircle of the storm. Spectral analysis of these images over 5km square

243

ocean cells were able to retrieve a direction that was nearly equal to the wind direction, as well as a mean wavelength (Katsaros et al., 2002). Fig. 1.b.16 (bottom) illustrates a histogram averaged over six such 2D spectra (6 km square) that shows a peak near 900 m. This value is close to that determined from ground-based WSR-88D Doppler radar observations in landfalling hurricanes (Morrison et al., 2005). Their results are supported by more recent higher-resolution portable Doppler radar observations in landfalling hurricanes (Losorlo et al., 2006a, b) and by Doppler on Wheels (DOW) high resolution portable Doppler radar observations in Hurricane Fran (Wurman and Winslow, 1998), and more recently in Hurricane Rita (Wurman et al., 2006). With these studies, the range of wavelengths of these linear features are well documented over land in landfalling hurricanes. Now, RADARSAT and ENVISAT SAR observations suggest they are also endemic to the hurricane wind field over the ocean.

Figure 1.b.16. RADARSAT SAR image (top) from right-front quadrant of Hurricane Fran, similar to that obtained for Hurricane Isidore, 2002. The horizontal scale is 200 km. Histogram (bottom) of wind streak wavelengths from analyzed RADARSAT image of Hurricane Isidore, 23 September, 2002. Arrow indicates peak in aircraft-derived spectrum in Fig. 1.b.22. RADARSAT image courtesy of Canadian Space Agency. To further emphasize possible boundary layer manifestation of these linear features, the along-wind component of the velocity co-spectrum was computed from gust probe data in Hurricane Isidore (2002) along a 75 km leg flown at 300 m, near the middle of the hurricane planetary boundary layer with a depth of 500 m (deduced from the depth of constant potential temperature layers measured by dropsondes). The first part of the leg was in the radial direction toward the eye and showed a significant peak in the co-spectrum near 900 m (Fig. 1.b.17), in close agreement with the scales from the SAR

244

histogram (Fig. 1.b.16, bottom). This peak disappeared when the aircraft turned and flew along wind.

Figure 1.b.17. Co-spectrum of vertical momentum flux (-u’w’) along a 120 m altitude radial flight leg into Hurricane Isidore, 22 September, 2002. The true air speed is 110 ms-1. The significance of these observations are that these linear features, and their possible major effects on air-sea fluxes as suggested by Morrison et al. (2005) and Foster (2005), are not currently modeled in any major hurricane coupled modeling effort, and may be an important factor in predictions of hurricane intensity change. This leads to the third major CBLAST finding to date, i.e. that secondary boundary layer circulations, while not a major thrust of the original CBLAST plan, are a significant factor in the hurricane boundary layer flow field and are a likewise significant factor in air-sea fluxes. 1.b.3.4: Key Results from the Air-Deployed Oceanographic Sensor Component of CBLAST Hurricane The 54 buoys and floats deployed into Hurricane Frances in 2004 yielded a wealth of information on ocean structure and structure changes induced by the hurricane within and below the ocean mixed layer: the first ever 4D ocean structure observations beneath a hurricane. Detailed measurements of the ocean and air-sea interface beneath hurricanes were made using several varieties of autonomous floats and drifters that were air-deployed ahead of the storm. The technology for these devices has matured rapidly in recent years so that they are now deployed in large numbers as part of the developing system for the Integrated Ocean Observing System (IOOS). For CBLAST, air-deployment systems were developed for existing platforms and they were equipped with new sensors to measure properties of the air-sea interface. The goal of these investigations was to understand the properties of the air-sea interface and upper ocean at wind speeds greater than 30 ms-1, to determine the associated air-sea fluxes and the effect of these on hurricane intensification.

245

The region of high winds beneath a hurricane is quite small: a few hundred km in diameter for the largest storms, much less for a typical storm. Accordingly, the probability of measuring high winds from an array of instruments prepositioned in “hurricane alley” is small. As with meteorological studies of hurricanes, a viable sampling plan must rely on real-time measurements of storm position, reliable forecasts of future storm tracks and the ability of aircraft to deploy sensors in or near an active storm based on this information. Accordingly, close cooperation was essential between the National Hurricane Center, which supplied the storm forecasts, the scientific team, who adapted the sampling array to these changing conditions, and the 53rd Air Force Reserve squadron, who deployed the instruments. Equally important was the use of UNOLS ships to recover the floats after the hurricane had passed.

Table 1.b.2. Oceanographic Platforms Deployed in CBLAST Hurricane

MiniMet ADOS EM-APEX Lagrangian SOLO

Type Drifter Drifter Float Float Float

Measurements SST

Air Pressure

Wind Speed

Wind

direction

Position

SST

Air Pressure

Wind Speed

Wind Direction

Temperature

0-120m

Position

Temperature

Salinity

Pressure

Velocity

Position

Temperature

Salinity

Pressure

Gas Tension

Oxygen

Position

Temperature

Salinity

Pressure

Oxygen

Sound

0-50kHz

Wave height

Position

Satellite Argos Argos Iridium Iridium Orbcomm

2003 Deployed 16 4 2

2004 Deployed 30 8 3 2 9

The five varieties of oceanographic instruments used in CBLAST Hurricane can be divided into two categories: drifters and floats. Details of each instrument type are shown in Table 1.b.2. Fig. 1.b.18 shows drawings of each instrument and a schematic of their operation in Hurricane Frances (2004). Drifters aim to follow the ocean current at 15m depth while measuring both near-surface atmospheric and upper-ocean properties. A small surface float supports a much larger drogue centered at 15m depth. The large drogue causes the drifter to nearly follow the horizontal water motion at approximately 15m. A transmitter in the surface drifter sends data to the ARGOS satellite system. The same signals are used to track the drifter. The standard drifter measurements are position and near-surface temperature. The CBLAST drifters carried additional sensors. Minimet drifters are also designed to estimate wind speed using the sound level at 8 kHz (Nystuen and Selsor, 1997) and wind direction using a vane on the surface float. Evaluation of the accuracy of this approach at hurricane wind speeds is still under way. ADOS drifters additionally measure the temperature profile to 100m depth.

246

Figure 1.b.18. Drawings of the three varieties of floats and a surface drifter as deployed into Hurricane Frances. Schematic depicts operations in Hurricane Frances (2004). The three varieties of floats are shown in Fig. 1.b.18. All floats operate by mechanically changing their volume, and thus their density, in order to control their depth. By making themselves light, they can profile to the surface thereby extending an antenna out of the water enabling them to obtain a GPS fix and relay data to and receive instructions from their shore-based operators. The EM-APEX floats (Fig 1.b.18, green lines) operated as profilers, continuously cycling while measuring temperature, salinity and velocity. Profiles extended from the surface to 200m with profiles to 500m every half inertial period. During the storm, the top of the profiles terminated at 50m. The Lagrangian floats (D’Asaro, 2003), profiled only before and after the storm (Fig. 1.b.18 black line). During the storm, they remained neutrally buoyant, following the three-dimensional motion of water parcels in the highly turbulent upper boundary layer. They measured temperature, salinity and gas concentration. The SOLO floats combined profiling of temperature, salinity and oxygen from the surface to approximately 200m (Fig. 1.b.18, blue line) while hovering at about 40m for a period of time during each dive interval to remotely measure surface waves and the depth of the bubble layer created by surface wave breaking using a compact sonar, and 0-50KHz ambient sound with a passive hydrophone. The floats were programmed to repeat its dive interval every 4 hours.

247

Figure 1.b.19. Hurricane Frances (2004) float and drifter array. Heavy gold line shows storm track, labeled by month/day, time (UTC). Colors indicate type of instrument: blue open circles are the drifters, red star are ARGO/SOLO floats, green diamonds are Lagrangian drifters and magenta squares are EMAPEX floats. Instrument tracks are plotted from deployment on JD 244 (31 Aug) to JD 246.5 (2 Sept, 12 UTC). Deployment position is indicated by black symbol. Initial deployments were made in Hurricane Isidore in 2002 and Fabian in 2003. The more extensive 2004 deployments ahead of Hurricane Frances will be described here. The array of floats and drifters (Fig. 1.b.19) was deployed from 20-23 UTC on Aug. 31, 2004 based on the Aug. 31, 03 UTC forecast for the 18 UTC, Sept. 1 storm position. The storm track passed just north of the shallow banks and islands north of Hispaniola. The array was therefore placed over deep water north of the storm track. The forecast proved to be extremely accurate, so that array elements passed under both the eye and maximum winds (60 ms-1) of the storm. The Hurricane Frances deployments clearly demonstrated the success of this new approach to measurement in hurricanes. The instruments were accurately targeted into the CAT 4 hurricane. All of the varieties of drifters and floats survived and worked successfully in this environment, with only minimal losses. Data was transmitted from the high wind region of the hurricane in nearly real time. The buoys and floats revealed a warm anticyclonic eddy directly in the path of Frances, which was flanked by cooler cyclonic features. Sea surface height anomaly maps from satellite altimetry for the North Atlantic3, such as the area along 30ºN east of northern Florida, sometimes referred to as the “Subtropical Convergence Zone”, indicates the presence of an eddy-rich ocean in the vicinity of the storm track. The time varying height of the sea surface was measured by the SOLO floats using an upward looking sonar compensated for the measured float depth. High frequency fluctuations in sea height yield

3 See Atlantic map at http://iwave.rsmas.miami.edu/heat and/or http://www.aoml.noaa.gov/phod/cyclone/data/

248

measurements of the surface waves. Maximum significant wave height exceeded 10m. Intense breaking of these large waves injects bubbles into the ocean. The resulting near-surface bubble layer plays a key role in gas flux across the air-sea interface as well as having an important dynamical effect by changing the bulk density of the near-surface layer. Bubbles are very efficient sound scatterers, so that the thickness of the near-surface bubble layer can be measured by the upward looking sonar. Its thickness increases approximately as wind speed cubed, reaching a maximum thickness of over 10m. These results are confirmed by Lagrangian float measurements of conductivity, which decreases in the upper 10m due to bubbles. Hurricanes draw their energy from the warm ocean waters. However, ocean mixing beneath a hurricane can significantly reduce sea surface temperatures from the pre-storm values. Fig. 1.b.20 shows the evolution of upper ocean potential density under the strongest winds of Hurricane Frances.

Figure 1.b.20. Evolution of the density structure of the upper ocean near the radius of maximum winds of Hurricane Frances. a) Wind speed and atmospheric pressure from HRD H*WIND analysis at the two Lagrangian floats. b) Potential density contours (kg m-3) in black, trajectories of Lagrangian floats in red and blue, measured depth of the mixed layer in magenta and estimated depth of the mixed layer from a vertical heat budget in yellow (dashed). It combines the vertical temperature profiles from an EM-APEX float with the nearby temperature measurements from the two Lagrangian floats. The EM-APEX floats also showed the evolution of the currents in the upper ocean (Sanford, et al., 2005). Fig 1.b.20 shows rapid deepening of the mixed layer and associated high shear across the thermocline. The strong wind and wave forcing directly generates turbulence in the upper 20-40m of the ocean.

249

Figure 1.b.21. EM-APEX float 1633 located 50 km to the right of storm track in highest winds. The upper two panels are of East (U) and North (V) velocity components versus depth and time with the 29 and 25ºC isotherms in bold with a contour interval of 0.5ºC. The center of the storm passed at approximately 1700 UTC on Sept 1. Ocean Heat Content (OHC) is shown for water warmer than 26ºC and shallower than 180 m. The bottom panel is ‘reduced shear’, a stability parameter, with the temperature contours superimposed. This quantity is related to the Richardson number, Ri=N2/S2, where N is the Brunt-Väisälä frequency and S is the vertical current shear. A necessary condition for shear instability is N2/S2 < 1/4. This instability criteria is re-written as S2 – 4N2 > 0 to indicate where mixing is possible. The quantity S2 – 4N2 is contoured in the bottom panel with the green, yellow and red colors indicating where the instability criteria is equaled or exceeded. The blue colors indicate stable conditions less than zero. The Lagrangian floats are advected by the large-eddy velocities of this turbulence, repeatedly cycling across the turbulent layer and thereby tracing its depth and intensity (red and blue lines). Turbulent velocities are 0.1 ms-1 rms at the height of the storm, with the strongest downward vertical velocities

250

reaching 0.3 ms-1. The turbulent layer extends to 50m at the height of the storm. However, the changes in temperature indicate that mixing extends to 120 m (magenta line). Measurements of shear by the EM-APEX floats show (Fig. 1.b.21) a nearly critical Richardson number down to 120m, indicating a key role for shear instability in this deeper mixing. The one dimensional heat budget requires even deeper mixing as shown by the yellow dashed line (Fig. 1.b.20, bottom panel). A more detailed analysis indicates that horizontal heat fluxes become important as the magenta and yellow-dashed lines diverge, indicating a transition of the boundary layer heat budget from vertical to three-dimensional. The net effect of this strong ocean mixing is to cool the ocean surface, potentially reducing the enthalpy flux to the hurricane. The combined data from the floats and drifters is used to map the amount of cooling in Fig. 1.b.22. Cooling is most intense to the right of the storm center, with a cold wake spreading outward behind this region. The leading edge of this wake forms an SST front approximately 50 km wide which moves with the storm. The eye of storm is at the edge of this front, so that cooling at the eye is only about 0.5˚C compared to the maximum of 2.5˚C in a crescent-shaped pattern in the storm’s right-rear quadrant, similar to that proposed by Black et al. (1988). SST gradients of up 2˚C exist across the inner 50 km of the storm, with a temperature range of about 27.5-30˚C. These data suggest that rather than specifying the SST at the hurricane inner core, it may be more useful to think in terms of the location of the SST front that exists beneath the core. Small changes in the location of this front relative to the core may have large effects on the enthalpy flux driving the storm.

Figure 1.b.22. SST decreases (C) beneath hurricane Frances (2003) in storm-centered coordinate system. White dots show storm-relative locations of float and drifter data. Storm motion is to left. Colors show mapped SST change from pre-storm value. Contours show wind speed in ms-1 from H*WIND analysis. Storm positions are in increments of one-quarter Julian Days (JD), or 6 hours, where JD 245 is Sept 1.

251

1.b.3.5 CBLAST-Hurricane Summary and Conclusions The CBLAST hurricane program has yielded an unprecedented data set for exploring the coupled atmosphere and ocean boundary layers during an active hurricane. Key results from the analysis effort to date have increased the range of air-sea flux measurements significantly, which have allowed drag and enthalpy exchange coefficients to be estimated in wind speeds to nearly hurricane force. The drag coefficients (CD’s) estimated from this work suggest a leveling off with wind speed near 22-23 ms-1, a considerably lower threshold than the 33 ms-1 value of Powell et al. (2003) and Donelan et al. (2004). This results in extrapolated CD’s for hurricane conditions above 33 ms-1 of under 0.002, slightly lower than the results of Powell et al. (2003) and Donelan et al. (2004) for the 30 to 40 ms-1 wind speed interval, which are approximately .0022. The Dalton number is constant with wind speed up to hurricane force with a value of 0.00118, in close agreement with modified HEXOS and COARE 3.0 estimates. This results in a CE/CD ratio of approximately 0.7, somewhat less than the Emanuel threshold of 0.75 for hurricane development. Laboratory results suggest that the inclusion of spray effects at high winds may cause this value to increase above one for major hurricanes (CAT 3 and above). Directional wave measurements made from the aircraft show distinctive characteristic as a function of storm-relative quadrant. Spectra range from tri-modal in the right-rear quadrant to bi-modal in the right front to unimodal in the left-front. The exchange coefficients appear independent of these wave spectral characteristics to within observational uncertainty. Breaking wave measurements will add to increased under standing of dissipation at the air-sea interface and contribute to a physical understanding of the variation of drag coefficient with wind speed, so important to model parameterizations. Boundary layer linear features (possible secondary circulations or ‘roll’ vortices) appear to characterize the boundary layer throughout the hurricane with their role in flux estimation yet to be determined. The drifter and buoy deployments in Hurricanes Fabian (2003) and Frances (2004) were unqualified successes yielding first time ever observations of the 4-dimensional evolution of the subsurface ocean structure concurrent with airborne atmospheric boundary layer observations. The development of the cold wake behind Frances, showing a crescent-shaped pattern of cooling in the near-storm environment, was well observed with maximum cooling of 2.5 ºC. Shear at the base of the ocean mixed layer was found to develop quickly beneath the hurricane and meet the Richardson Number criteria of 1/4 for onset of turbulent mixing. Future work will focus on integrating existing and anticipated results on air-sea flux parameterization into high-resolution (1.7 km), coupled hurricane models such as Chen et al., 2006, as well as the new HWRF model, scheduled to become operational at the NOAA Environmental Prediction Center (EMC) during the 2006 hurricane season. EMC will also be examining the benefits of assimilating the profiling float data into their operational models to assess the value of deploying similar instruments in future storms to improve intensity predictions. Similarly, efforts are being made to integrate CBLAST results into the Navy NOGAPS and COAMPS models. Special efforts will begin to assess the impact of the new air-sea parameterization schemes on hurricane intensity. We are at a unique point in history where airborne advanced technology is able to meet the requirements of the new generation of advanced coupled models for input and validation data in the hurricane, at the air-sea interface and in the ocean. Advances in hurricane computer modeling and observational technology are symbiotic. Continued investment in this effort now will produce large dividends at low risk for future improved hurricane intensity and track prediction.

252

Acknowledgements. The authors and team of CBLAST investigators wish to thank the Office of Naval Research for supporting this work as a Departmental Research Initiative, award numbers N00014-01-F-009 (Black, Drennan, French), N00014-01-1-0162 (Emanuel, Black), N00014-00-1-0894 (Terrill, Melville), N00014-00-1-0893 (D’Asaro), N00014-02-1-0401 (Niiler) and N00014-01-F-0052 (Walsh). The United States Weather Research Program (USWRP) also supported Drennan with grant number NA17RJ1226. Walsh was also supported by the NASA Physical Oceanography Program. EM-APEX floats were developed with an ONR SBIR grant to Webb Research, Inc. We thank the NOAA Office of Oceanic and Atmospheric Research (OAR) for P3 flight hour and expendable dropsonde support for three years and to USWRP for CBLAST analysis support from 2005 to the present. Special thanks is due to ONR CBLAST Program managers Scott Sandgathe , Simon Chang and Carl Friehe for actively helping to guide the program. Thanks is also extended to Theresa Paluszkiewicz, Ron Ferek, Linwood Vincent and Steve Tracton for ONR management support. References

Andreas, E. L. and K. A. Emanuel, 2001: Effects of Sea Spray on Tropical Cyclone Intensity. J. Atmos. Sci., 58, 3741–3751. Bell, M. M. and M. T. Montgomery, 2006: Observed structure, evolution and potential intensity of category five Hurricane Isabel (2003) from 12 – 14 September. Submitted to Mon. Wea. Rev. Black, P. G., E. A. D’Asaro, W. M. Drennan, J. R. French, P. P. Niiler, T. B. Sanford, E. J. Terrill, E. J. Walsh and J. A. Zhang, 2006: Air-Sea Exchange in Hurricanes: Synthesis of Observations from the Coupled Boundary Layer Air-Sea Transfer Experiment. Accepted Bull. Amer. Meteorol. Soc. Black, P. G., R. L. Elsberry and L. K. Shay, 1988: Airborne surveys of Ocean Current and temperature perturbations induced by hurricanes. Adv. Underwater Technol., Ocean Sci. Engineer., 16, 51-58. Chen, S. S., J. F. Price, W. Zhao, M. A. Donelan, and E. J. Walsh, 2006: The CBLAST-Hurricane program and the next generation fully coupled atmosphere-wave-ocean models for hurricane research and prediction. Bull. Amer. Meteorol. Soc., in press.. Chen, S. S., 2006: Overview of RAINEX modeling of 2005 hurricanes. The 27

th Conf. on Hurricanes

and Tropical Meteorology. 23-27 April 2006, Monterey, CA, Amer. Meteor. Soc. 12A.2.

D'Asaro, E.A., 2003: Performance of autonomous Lagrangian floats. J. Atmos. Oceanic Technol., 20, 896-911. DeCosmo, J., K. B. Katsaros, S. D. Smith, R. J. Anderson, W. A. Oost, K. Bumke, and H. Chadwick, 1996: Air-sea exchange of water vapor and sensible heat: The humidity exchange over the sea (HEXOS) results. J. Geophys. Res., 101 (C5), 12001-12016. De Leeuw, G., 1990: Profiling of aerosol concentrations, particle size distributions and relative humidity in the atmospheric surface layer over the North Sea. Tellus, 42B, 342-354. DeMaria, M., M. Mainelli, L.K. Shay, J.A. Knaff, and J. Kaplan, 2005: Further Improvements to the Statistical Hurricane Intensity Prediction Scheme (SHIPS), Wea. Forecasting, 20, 531-543. Donelan, M. A., 1990: Air-Sea Interaction, The Sea: Ocean Engineering Science, B. LeMéhauté and D. Hanes, Eds., Vol. 9, John Wiley and Sons, New York, pp. 239-292.

253

Donelan, M. A., B. K. Haus, N. Reul, W. J. Plant, M. Stianssnie, H. C. Graber, O. B. Brown, E. S. Saltzman, 2004: On the limiting aerodynamic roughness of the ocean in very strong winds. Geophys. Res. Lett., 31, L18306. Drennan, W. M., J. Zhang, J. R. French, C. McCormick and P. G. Black, 2006: Turbulent fluxes in the hurricane boundary layer, II. Latent heat flux, Accepted, J. Atmos. Sci. Edson, J., T. Crawford, J. Crescenti, T. Farrar, J. French, N. Frew, G. Gerbi, C. Helmis, T. Hristov, D. Khelif, A. Jessup, H. Jonsson, M. Li, L. Mahrt, W. McGillis, A. Plueddemann, L. Shen, E. Skyllingstad, T. Stanton, P. Sullivan, J. Sun, J. Trowbridge, D. Vickers, S. Wang, Q. Wang, R. Weller, J. Wilkin, D. Yu, and C. Zappa, 2006: The Coupled Boundary Layers and Air-Sea Transfer Experiment in Low Winds (CBLAST-LOW). Accepted Bull. Amer. Meteorol. Soc.. Emanuel, K. A., 1986: An Air-Sea Interaction Theory for Tropical Cyclones. Part I: Steady-State Maintenance. J. Atmos. Sci., 43, 585–605. Emanuel, K. A., 1995: Sensitivity of tropical cyclones to surface exchange coefficients and a revised steady-state model incorporating eye dynamics. J. Atmos. Sci., 52, 3969-3976. Emanuel, K. A., 2005: Increasing destructiveness of tropical cyclones over the past 30 years. Nature, 436, 686-688. Esteban-Fernandez, D., S. Frazier, J. Carswell, P. Chang, P. Black and F. Marks, 2004: 3-D atmospheric boundary layer wind fields from Hurricanes Fabian and Isabel. 26th Conf. on Hurr. and Tropical Meteorol., AMS, Miami Beach, FL. Fairall, C. W., W. Asher, M. Banner, and W. Peirson, 2006: Investigation of the physical scaling of sea spray spume droplet production. J. Geophys. Res., in preparation. Fairall, C. W., J. E. Hare, and A A. Grachev, 2004: Sea Spray Droplet Measurements in Hurricanes Fabian and Isabel. Proc. 26th Conf. Hurricanes and Tropical Meteorlogy. AMS Miami, FL, 1-5 May. Fairall, C. W., E. F. Bradley, J. E. Hare, A. A. Grachev and J. B. Edson, 2003: Bulk parameterization of air-sea fluxes: updates and verification for the COARE algorithm. J. Climate, 16, 571-591. Foster, R. C., 2005: Why rolls are prevalent in the hurricane boundary layer. J. Atmos. Sci., 62, 2647-2661. Franklin, J.L., C. J. McAdie, and M. B. Lawrence. 2003: Trends in Track Forecasting for Tropical Cyclones Threatening the United States, 1970–2001. Bull. Amer. Meteor. Soc., 84, 1197–1203. French, J.R., W.M. Drennan, J. Zhang, and P.G. Black, 2006: Turbulent Fluxes in the Hurricane Boundary Layer, I. Momentum Flux. Accepted J. Atmos. Sci. Goldenberg, S.B., C.W. Landsea, A. M. Mestas-NuÑez, W. M. Gray, 2001: The Recent Increase in Atlantic Hurricane Activity: Causes and Implications. Science 293, 474-478. Hebert, P. J., 1976: Atlantic hurricane season of 1975. Mon. Wea. Rev., 104, 453-475. Hildebrand, P. H., W.-C. Lee, C. A. Walther, C. Frush, M. Randall, E. Loew, R. Neitzel, R. Parsons, J. Testud, F. Baudin and A. LeCornec, 1996: The ELDORA/ASTRAIA Airborne Doppler Weather Radar: High-Resolution Observations from TOGA COARE. Bull. Amer. Meteor. Soc., 77, 213–232. Houze, R. A., S. S. Chen, and co-authors, 2006: The Hurricane Rainband and Intensity Change

254

Experiment (RAINEX): Observations and modeling of Hurricanes Katrina, Ophelia, and Rita (2005). Bull. Amer. Meteor. Soc, 87, in press.

Katsaros, K. B., P. W. Vachon, W. T. Liu, and P. G. Black, 2002: Microwave remote sensing of tropical cyclones from space. J. Oceanogr., 58, 137-151. Kepert, J. D., C. W. Fairall, and J.-W. Bao, 1999: Modeling the interaction between the atmospheric boundary layer and evaporating sea spray droplets. Air-Sea Fluxes: Momentum, Heat, and Mass Exchange, G. L. Geernaert, Editor, Kluwer, Dordrecht, 363-409. Kleiss, J. M., W. K. Melville, J. R. Lasswell, P. Matusov and E. Terrill, 2004: Breaking waves in hurricanes Isabel and Fabian. 26th Conf. on Hurr. and Tropical Meteorol., AMS, Miami Beach, FL. Large, W. G., and S. Pond, 1981: Open ocean momentum flux measurements in moderate to strong winds. J. Phys. Oceanogr., 11, 324-336. Landsea, C.W., 2005: Hurricanes and global warming. Nature, 438, E11-E12. Lorsolo, S. and J. R. Schroeder, 2006a: Small scale features observed in the boundary layer of Hurricanes Isabel (2003) and Frances (2004). Proc. Air-Sea Interaction Conf., AMS, Atlanta, GA. Lorsolo, S. and J. R. Schroeder, 2006b: Tower and Doppler radar observations from the boundary layers of Hurricanes Isabel (2003) and Frances (2004). Proc. 27th Conference on Hurricanes and Tropical Meteorology, AMS, Monterey, CA. Montgomery, M. T., M. M. Bell, S.Aberson, and M. Black, 2006: Hurricane Isabel (2003): New insights into the physics of intense storms. Part I. Mean vortex structure and maximum intensity estimate. In press, Bull. Amer. Meteorol. Soc. Morrison, I., S. Businger, F. Marks, P. Dodge and J. Businger, 2005: An observational case for the prevalence of roll vortices in the hurricane boundary layer, J. Atmos. Sci. 62, 2662-2673 Moss, M. S., 1978: Low-Level Turbulence Structure in the Vicinity of a Hurricane, Mon. Wea. Rev., 106, 841–849. Moss, M. S. and F. J. Merceret, 1977: A Comparison of Velocity Spectra from Hot Film Anemometer and Gust-Probe Measurements, J. Appl. Meteorol.,16, 319–320. Moss, M. S. and F. J. Merceret, 1976: A Note on Several Low-Layer Features of Hurricane Eloise (1975), Mon. Wea. Rev., 104, 967–971. Nystuen, J. A. and H. D. Selsor, 1997: Weather classification using passive acoustic drifters. J. Atmos. and Oceanic Tech., 14, 656-666. Ortt, D, and S. S. Chen, 2006: Rainbands and secondary eyewall formation as observed in RAINEX. 27

th Conf. on Hurricanes and Tropical Meteorology. 23-27 April 2006, Monterey, CA, Amer. Meteor.

Soc. 12A.5.

Powell, M. D., and S. H. Houston, 1996: Hurricane Andrew's Landfall in South Florida. Part II: Surface Wind Fields and Potential Real-time Applications. Weather. Forecast., 11, 329-349. Powell, M. D., S. H. Houston, L. R. Amat, and N Morisseau-Leroy, 1998: The HRD real-time hurricane wind analysis system. J.Wind Engineer. and Indust. Aerodyn. 77&78, 53-64.

255

Powell, M. D., and S. H. Houston, 1998: Surface wind fields of 1995 Hurricanes Erin, Opal, Luis, Marilyn, and Roxanne at landfall. Mon Wea. Rev., 126, 1259-1273. Powell, M. D., P. J. Vickery, and T. A. Reinhold, 2003: Reduced drag coefficient for high wind speeds in tropical cyclones. Nature, 422, 279-283. Sanford, T. B., J. H. Dunlap, J. A. Carlson, D. C. Webb and J. B. Griton, 2005: Autonomous Velocity and Density Profiler: EM-APEX. Proceedings of the IEEE/OES Eighth Working Conference on Current Measurement Technology, IEEE Cat No. 05CH37650, ISBN: 0-7803-8989-1, pp. 152-156 Smith, S. D., 1980: Wind stress and heat flux over the ocean in gale force winds. J. Phys. Oceanogr., 10, 709-726. Smith, S.D., R.J. Anderson, W.A. Oost, C. Kraan, N. Maat, J. DeCosmo, K.B. Katsaros, K.L. Davidson, K. Bumke, L. Hasse and H.M. Chadwick, 1992: Sea surface wind stress and drag coefficients: the HEXOS results. Bound. Layer Meteorol. 60, 109-142. Uhlhorn, E. W., and P. G. Black, 2003: Verification of remotely sensed sea surface winds in hurricanes. J. Atmos. Ocean. Technol., 20, 99-116. Uhlhorn, E. W., P. G. Black, J. L. Franklin, M. Goodberlet, J. Carswell, and A.S. Goldstein, 2006: Hurricane surface wind measurement from an operational Stepped Frequency Microwave Radiometer. Submitted J. Atmos. Oceanic Technol. Webster, P. J., G. J. Holland, J. A. Curry, H.-R. Chang, 2005: Changes in Tropical Cyclone Number, Duration, and Intensity in a Warming Environment. Science, 309, 1844-1846. Willoughby, H. E., 1988: The dynamics of the tropical hurricane core. Aust. Meteor. Mag., 36, 183-191. Wright, C. W., E. J. Walsh, D. Vandemark, W. B. Krabill, S. H. Houston, M. D. Powell, P. G. Black, and F. D. Marks, 2001: Hurricane directional wave spectrum spatial variation in the open ocean. J. Phys. Oceanogr., 31, 2472-2488. Wurman, J. and J. Winslow, 1998: Intense sub-kilometer scale boundary layer rolls observed in Hurricane Fran. Science, 280, 555-557. Wurman, J, C. Alexander, P. Robinson and F. Masters, 2006: Preliminary comparisons of DOW and insitu measurements in Hurricane Rita. Proc. 27th Conference on Hurricanes and Tropical Meteorology, AMS, Monterey, CA. Yelland, M.J., B.I. Moat, P.K. Taylor, R.W. Pascal, J. Hutchings and V.C. Cornell, 1998: Wind stress measurements from the open ocean corrected for airflow distortion by the ship. J. Phys. Oceanogr. 28, 1511-1526. Zhao, W., and S. S. Chen, 2006: Effects of surface waves and upper ocean on hurricane structure and intensity in a fully coupled model. The 27

th Conf. on Hurricanes and Tropical Meteorology. 23-27 April

2006, Monterey, CA, Amer. Meteor. Soc. 5C.7.

256

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES Topic 2 : Tropical Cyclone Formation and Extratropical Transition Topic Chair: Patrick A. Harr Naval Postgraduate School Monterey, CA 93943-5114 USA E-mail: paharr@nps.edu FAX: 01.831.656.3061 2.0.0 Introduction In this section, summaries of recent progress in observations, forecasting, and research of tropical cyclone formation and extratropical transition are provided. Furthermore, requirements that remain for advancements in these topics are identified. During the period since IWTC-V, there have been improvements in observational capability, new research results into important physical processes, and forecast improvements with respect to tropical cyclone formation and extratropical transition. However, further requirements have also been identified. A partial list of these requirements includes improved definitions of formation and extratropical transition, validating theories of tropical cyclone formation, identifying the role(s) of the large-scale environment in tropical cyclone formation and extratropical transition, sensitivities in formation and extratropical transition to important physical processes, and the impact of an extratropical transition event on the downstream midlatitude circulation. The tropical cyclone formation and extratropical transition remain challenging complex problems because they involve interactions among physical processes that vary over multiple space and time scales. Therefore, progress in all aspects of tropical cyclone formation and extratropical transition will be realized via a collaboration of forecasters, observationalists, modelers, and theoreticians. In this summary, the external and internal influences in tropical cyclone formation are examined. Then, the forecast challenges associated with tropical cyclone formation are examined. The summary of extratropical transition is based on observational and forecast challenges followed by factors related to physical processes and downstream impacts. 2.0.1 External Influences on Formation For some time, the large-scale, climatological conditions favorable for the formation of tropical cyclones have been known as a combination of dynamical and thermodynamic factors. These include a sea-surface temperature above 26oC, a deep ocean mixed layer, cyclonic low-level vorticity, and weak vertical wind shear. Furthermore, there is often a region of organized deep convection that exists in a region of low- and mid-level moisture. Over recent years, there has been increased attention on the roles of tropical wave activity on the organization of deep convection as a pre-cursor to tropical cyclone formation. While tropical easterly waves over the North Atlantic and eastern North Pacific have long been identified as pre-cursor disturbances to tropical cyclone formation over those regions, there has been increased focus on tropical wave activity over additional ocean basins that contain tropical cyclones. Some of this focus is directly attributable to the availability of new data sources that allow identification of wave characteristics over a variety of spatial and temporal scales.

257

Overall, tropical wave activity may be placed in the context of basic modes of circulation associated with special dynamic and thermodynamic characteristics over equatorial latitudes. Often, the basic structural features of the basic tropical wave modes are not conducive or consistent with large-scale characteristics known to be favorable for tropical cyclone formation (i.e., latitude, rotational flow), However, over several ocean basins, interactions between tropical waves and large-scale, basic-state flow characteristics may alter the wave structures such that its potential for evolution into a tropical cyclone becomes more favorable. Recent research has reinforced the roles of basic tropical circulation modes on climatological factors such as rainfall, cloud distribution, and tropical cyclone formation. Beyond the roles of individual modes of tropical circulation, the interactions among modes are important for altering circulation characteristics that may be favorable for tropical cyclone formation. Furthermore, the interaction among modes may be responsible for changes in storm frequency that could vary from synoptic time scales to intraseasonal time scales. While tropical wave activity has been investigated as an influence on tropical cyclone formation characteristics, it is important to examine factors that impact the evolution of a circulation from a wave-like feature to a tropical cyclone vortex. Many of these factors are external to the wave activity. These include vertical wind shear, and increased baroclinicity that may result from trough intrusions from the midlatitudes to the subtropics or tropics. Finally, there are indications that the relationships among large-scale factors and tropical cyclone formation may change as climate changes. Therefore, the relative importance of individual dynamic and thermodynamic factors may change over future years. Because of the availability of several new data sets, which include reanalysis fields and remotely-sensed parameters such as outgoing longwave radiation, it is recommended that studies make use of these data sets to assess the impacts of variability in tropical circulation modes on tropical cyclone formation. It is also recommended that the impacts of external factors (i.e., vertical wind shear) be assessed relative to tropical cyclone formation from basic tropical modes of circulation. If clear relationships between tropical cyclone formation and tropical circulation modes exist, then it is recommended that the potential for prediction of tropical cyclone formation be explored by utilizing statistical relationships between known tropical circulation characteristics (i.e., periodicity, vertical variation in amplitude) and tropical cyclone formation. Relationships that would lead to statistical predictive skill require identification of many factors such as the role of basin-scale circulation changes, impacts of vertical wind shear, and sea-surface temperature variations. 2.0.2 Internal Influences on Tropical Cyclone Formation For a variety of reasons, the tropical cyclone formation has remained one of the least understood phases of the tropical cyclone lifecycle. This is typically related to the lack of conventional observations over oceanic regions in which tropical cyclones form. However, increased satellite sensors and improved analysis techniques of remotely-sensed data have provided better representation of the tropical cyclone formation process. The coverage of satellite data in conjunction with some specialized observation campaigns, has led to two primary theories for the organization of a precursor tropical disturbance into a tropical cyclone. These theories of tropical cyclone formation have been advanced using detailed modeling studies that allow examination of the sensitivity of tropical cyclone formation to a variety of mechanisms. Since IWTC-V, the primary theories of tropical cyclone formation have been argued from the viewpoint of bottom-up or top-down development of a low-level cyclonic vorticity maximum. In the bottom-up scenario, pre-existing low-level cyclonic vorticity is increased due to convergence and stretching

258

associated with regions of deep convection. Observational evidence exists that define vortex intensification following periods of intense deep convection. In the top-down scenario, a precursor tropical disturbance may contain one or more mesoscale convective systems (MCSs) with an accompanying mesoscale convective vortex (MCV). Merger of MCSs within the original disturbance results in a more intense MCV that may eventually extend downward to increase the cyclonic vorticity at low levels. There are crucial internal elements to each theory of tropical cyclone formation. For example, the bottom-up theory requires some pre-existing cyclonic vorticity at low levels while the top-down scenario requires sustained precipitation to saturate the relatively dry and cold lower levels beneath the stratiform region of the MCS/MCV complex. While the definition of tropical cyclone formation generally focuses on the presence of low-level cyclonic vorticity and a warm-core vertical structure, the differences in the bottom-up and top-down theories of formation often lead differences in the development of these features. Therefore, a specific definition of tropical cyclone formation is still required, which was also identified as a requirement during IWTC-V. Research of internal influences on tropical cyclone formation should be guided by several remaining over-arching questions on key physical mechanisms. These mechanisms include the requirement of a saturated or nearly saturated profile associated with a downdraft-free convective environment. Furthermore, it is recommended that a widely-accepted definition of tropical cyclone formation that concentrates on the evolutionary process. There is a continued requirement for detailed investigations via high-resolution numerical modeling of the processes that lead to increases in low-level vorticity in tropical disturbances. Finally, null cases or non-developing tropical disturbances should be investigated to determine sensitivities to various physical processes. 2.0.3 Operational Forecasting of Tropical Cyclone Formation Many factors contribute to challenges in operational forecasting of tropical cyclone formation. Because tropical cyclones tend to form in regions with relatively few conventional observations, forecasters must rely heavily on satellite data and global operational numerical weather prediction models. Furthermore, there are specialized issues associated with the operational forecasting of tropical cyclone formation in each ocean basin in which tropical cyclones form. For example, over the southern West Pacific and eastern South Indian Ocean tropical cyclones may form very near coastal locations. Therefore, there is very little time between formation and a potential landfall. Even if the tropical cyclone does not make landfall, formation of a tropical cyclone near the coast of Eastern or Western Australia is often associated with periods of heavy rainfall. Over recent years since IWTC V, much emphasis has been placed on identifying the ability of global operational numerical prediction models. Although a large amount of variation in skill among the operational models exists, it is generally recognized that skill in prediction of tropical cyclone formation has increased. Although special cases of rapid cyclogenesis or cyclogenesis in high vertical wind shear conditions are not forecast well, many factors seem to contribute to the increase in skill. Tropical cyclone formation in conjunction with well defined large-scale conditions is one instance when numerical forecasts exhibit some skill. Specific favorable large-scale conditions include the presence of the convectively-active phase of the Madden-Julian Oscillation (MJO). Although it is generally agreed that operational global models do not exhibit predictive skill associated with the MJO, in basins such as the western North Pacific and regions of the Australian monsoon trough, well represented MJO conditions in the initial conditions do seem to persist long enough into forecast sequences to reflect favorable conditions for tropical cyclone formation. There has been some documented success of predictive skill associated with tropical cyclone

259

formation during periods of other favorable large-scale circulation modes (i.e., equatorial Rossby waves). In particular, several instances of multiple outbreaks of tropical cyclones related to Rossby wave dispersion of pre-existing tropical cyclones have been documented. There is a need to transition conceptual models that develop from validated theories of tropical cyclone formation to operational practice. This transition should utilize new satellite data sources and increased skill from numerical prediction models. Tools for systematic evaluation of the distribution and evolution of deep convection in pre-cursor circulations based on satellite data should be developed. It is recommended that the skill associated with each global operational numerical model be evaluated with respect to forecasts tropical cyclone formation in each ocean basin in which tropical cyclones form. Knowledge of the model performance traits can alert forecasters to increase monitoring of disturbances likely to intensify into tropical cyclones. Finally, it is recommended that conceptual models of factors related to basic tropical modes of circulation be organized for operational use. 2.0.4 Observing and Forecasting of Extratropical Transition In recent years, the difficulty in operational forecasting the extratropical transition of tropical cyclones has been highlighted relative to diagnostic analysis of observations and numerical prediction. Often, extratropical transition is associated with maintenance of tropical cyclone force winds, precipitation, and ocean waves far into the midlatitudes. Since official forecasts of the tropical cyclone may have been terminated, it is often difficult to convey the continued threat of the damaging weather elements to the public. Therefore, a requirement for a precise definition of extratropical transition is required such that the continued threat may be conveyed to the general public. Furthermore, the definition should encompass the needs of operational and research communities for a common framework on which to examine the physical processes that occur during extratropical transition. In recent years, several cases of extratropical transition have been observed with aircraft such that new data sets have been available for analysis of the complex physical processes that occur during extratropical transition. While these studies have provided new insights into changes of the remnant tropical cyclone structure there are still requirements for increased forecast skill associated with many of the impacts of extratropical transition such as the expansion of the surface wind distribution, the distribution of heavy precipitation, occurrence of tornadoes, and the generation of extreme ocean wave conditions. One particular forecast issue associated with extratropical transition has been identification of the timing of the extratropical transition process and the likely structural characteristics associated with the extratropical cyclones that result from the extratropical transition process. Operational forecasting of these factors has improved with the use of the cyclone phase space for summary and display of analyzed and forecast structural characteristics of a tropical cyclone as it proceeds through the extratropical transition process. Furthermore, detailed numerical simulations of the extratropical transition process have been instrumental in identifying sensitivities to tropical cyclone characteristics and the midlatitude circulation into which the decaying tropical cyclone is moving. In conjunction with numerical simulations, diagnostic analyses of extratropical transition events have led to specification of factors important for increasing forecast skill of many attributes of an extratropical transition event. While some diagnostic analyses have been performed on conventional atmospheric fields, others have been based on the cyclone phase space, which provides a compact framework for assessing the dynamic and thermodynamic characteristics of the extratropical transition process. Specific recommendations for improvements in forecasting of extratropical transition events revolve around the timing of the evolutionary processes and the sensitivity of the process to various physical characteristics. These improvements are ultimately connected to observations of critical processes such that appropriate diagnosis of the extratropical transition may be conducted and the processes may be accurately depicted in the initial conditions of operational numerical forecast models. Critical

260

to these improvements is the increased utility of satellite data and in particular microwave imagery from polar-orbiting platforms that will identify thermodynamic and moisture distributions during the extratropical transition event. Ultimately, the development of a satellite-based extratropical transition diagnostic should be developed. 2.0.5 Physical Processes and Downstream Impacts of Extratropical Transition Since IWTC-V, one of the primary realizations associated with extratropical transition has been the importance of extratropical transition on influencing the multi-scale dynamics associated with the midlatitude circulation far downstream of the extratropical transition event. The downstream impacts may be related to a variety of physical processes associated with the extratropical transition and the midlatitude circulation in which the extratropical transition is occurring. Physical processes associated with the extratropical transition involve the transport of anomalous amounts of moisture and heat into the midlatitudes in conjunction with the presence of a positive potential vorticity (PV) anomaly. Therefore, diabatic processes often play a critical role in altering the distribution of PV and may modulate the impact of the PV anomaly associated with the decaying tropical cyclone on the pre-existing midlatitude PV distribution. Additionally, the transport of heat and moisture has important implications on boundary layer contributions to the re-intensification of the decaying tropical cyclone as an extratropical cyclone. Often the movement of the tropical cyclone into the midlatitudes imposes significant perturbation to the midlatitude flow that may rapidly extend downstream and to a lesser degree upstream of the location of the extratropical transition event. The large amount of variability in the downstream response to an extratropical transition suggests that there are important sensitivities to a variety of physical mechanisms associated with the forcing on the midlatitude flow. These mechanisms may include basic baroclinic energy conversions, forcing of diabatically-forced Rossby wave-like circulations, and enhancements to downstream jet streaks. All of these factors would exhibit sensitivity to a variety of interactions with a decaying tropical cyclone. Finally, use of ensemble prediction systems indicate that the downstream forcing due to an extratropical transition is typically associated with reduced predictability. Not only is extratropical transition associated with severe local weather conditions, the impact of an extratropical transition on the midlatitude flow has been shown to be directly linked to high-impact weather events far downstream of the original extratropical transition event. Therefore, there is a strong requirement to assess the extent of the sensitivity to various physical mechanisms during extratropical transition. These mechanisms include the impact of sensible and latent heat fluxes, precipitation distribution, frontogenesis, and baroclinic energy conversions. Finally, the character of the downstream response to extratropical transition should be placed in the framework of the mean environmental conditions (i.e., baroclinic wave guides) across the entire ocean basin in which the extratropical transition is occurring.

261

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES Topic 2.1 : External Influences on Formation Rapporteur: William M. Frank The Pennsylvania State University 503 Walker Building University Park, PA, 16827, USA E-mail: Frank@ems.psu.edu Fax: 814.865.0470 Working Group: W. M. Frank, G. J. Holland, P. Klotzbach, J. L. McBride, P. E. Roundy, with additional contributions from J. Molinari. Abstract: 2.1.1: Introduction Understanding how tropical cyclones form and how the structure and intensity of these storms varies under the influence of larger-scale circulations is fundamental to improving forecasts and response plans in the current and future climate regimes. The climatological conditions present when they form have been well known since Gray (1968, 1979) and were summarized by Briegel and Frank (1997) and others: sea surface temperatures above about 26.5C with a deep ocean mixed layer; a positive low-level vorticity anomaly; weak and preferably easterly vertical wind shear; and a region of organized deep convection with moistened lower and middle levels. However, individual storms form infrequently and sporadically within large areas of favorable environmental conditions due to the effects of local flow perturbations. The key to understanding tropical cyclogenesis lies in determining how the full range of tropical weather phenomena collectively produce the sufficient local conditions to form a storm. Further, it is essential to understand the relationships between the occurrence of genesis conditions locally and the basin-scale state of the atmosphere and ocean. Most of the tropical cyclogenesis research since the previous IWTC has fallen in two general areas. One of these involves studies of the relationships between tropical wave activity, and the other is a look at changes in storm formation frequency within cyclone basins on time scales ranging from interannual to multi-decadal. 2.1.2 Tropical Waves and Cyclogenesis

The most active area of research on the genesis of individual storms during the past few years has been the role of tropical waves in cyclogenesis. Frank and Roundy (2006) argue that the favorable anomalies that produce most of the world’s tropical cyclones are organized by one or more tropical waves, often as they interact with a monsoon trough and/or each other. It has been known for decades that the tropical atmosphere is perturbed by a variety of wave types, several of which are unique to the equatorial wave duct. These waves were analyzed analytically by Matsuno (1966), Zhang and Webster (1989), and others. However, they have always been difficult to study due to the paucity of conventional data over the tropical oceans. Spectral analyses and composite studies (e.g. –Reed et all, 1977) revealed some aspects of their structures in the real atmosphere, but it was

262

not until improved models and data assimilation techniques were developed that the waves were resolved well enough in global analysis fields to permit detailed studies of their structures and behavior. Recent studies of tropical wave characteristics include Hendon and Liebmann (1991); Madden and Julian (1994); Takayabu (1994); Wheeler and Kiladis (1999); Wheeler and Webster (2000); Straub and Kiladis (2002); and Kiladis et al. (2005). Roundy and Frank (2004 a,b,c) studied the climatology and properties of tropical waves on a global basis, and they analyzed the net effects of the various wave types upon rainfall variability in the tropics. They found that the waves play a larger role in modulating tropical rainfall patterns than had had been generally recognized, with individual wave types providing as much as 40% of the total rainfall variance in some regions. They also showed that interactions between wave types and between the waves and the background flow and topography are important features in the behavior and effects of the waves. Since the waves modulate the large-scale vertical velocity, vorticity, and vertical shear patterns, they can also have potentially large effects on tropical cyclone formation and structure. There have been several recent observational case studies of the influence of tropical waves on tropical cyclogenesis – e.g.: Landsea (1993), Thorncroft and Hodges (2001), and others have studied the genesis of hurricanes in the Atlantic resulting from African waves. Molinari and Vollaro (2000), Maloney and Hartmann (2001), Liebmann et al. (1994), and others have showed effects of the MJO (Madden-Julian Oscillation) on North Pacific tropical cyclogenesis. Dickinson and Molinari (2002) analyzed events of cyclogenesis in the Northwest Pacific associated with Mixed-Rossby-Gravity waves, while Bessafi and Wheeler (2005) and Hall et al. (2001) analyzed the relationships between the MJO, other wave types, and cyclogenesis over the southern Indian Ocean and in the Australian region, respectively. Frank and Roundy (2006) analyzed relationships between tropical cyclone formation and tropical wave activity in each of the six global tropical cyclone basins. Using statistical and compositing techniques applied to spectrally filtered outgoing longwave radiation (OLR) data and reanalysis wind fields they were able to show that most tropical cyclones form in the convectively active portions of tropical waves, and that the phase relationship between the genesis location and each wave type is the same in every cyclone basin. All of the wave types except for Kelvin waves clearly played significant roles in tropical cyclone formation. The higher frequency waves were most important in the N. Hemisphere, while the MJO dominated S. Hemisphere genesis. Based on the combined results of the above studies, it is clear that tropical waves have significant effects upon tropical cyclogenesis, and they almost certainly have significant effects on mature cyclones as well. It is time to examine in a quantitative fashion how the waves produce these effects. In order to do this it is necessary to combine observations of individual storms and their environments, knowledge of wave structures determined from the previously described statistical and compositing studies of waves, and high-resolution numerical simulations. 2.1.3 Other Factors Affecting Genesis Several recent studies have examined environmental flow conditions at the time and place of genesis and the physical interactions between the large-scale flow and cyclone formation. Davis and Bosart (2003) examined baroclinically induced development of tropical cyclones and found that a sharp reduction in the vertical shear that resulted from diabatic heating was essential to the formation. Karyampudi and Pierce (2002) showed that several different synoptic-scale processes associated with the Saharan Air Layer can interact with midlevel wave vortices to form storms in the Atlantic. Molinari and Corbisiero (2004) analyzed the development of a hurricane core in a sheared environment. A modeling study by Bister (2001) showed that strengthened convection on the periphery of a developing storm tended to retard the core development.

263

There were also studies of new statistical techniques for forecasting genesis. Hennon et al. (2003,2005) explored improvements in statistical forecasting of tropical cyclogenesis in the North Atlantic using large-scale data and neural network techniques. McDonnell and Holbrook (2004) developed a Poisson regression model for forecasting genesis in the Southwest Pacific, while DeMaria et al. (2001) developed a genesis parameter for the North Atlantic. 2.1.4 Relationships Between Genesis and Basin-Scale Variations Recent studies by Webster et al. (2005), Emanuel (2005), and Mann and Emanuel (2006) argue that tropical cyclones appear to have increased in number and intensity during the last century or so. They suggest that these changes could be related to small increases in the mean sea surface temperatures (SST) that have been observed during that period. While the problem of tropical cyclones in future climates is the focus of another working group, one aspect of the problem is important to this one. What are the relationships between long-term basin-scale circulation changes and the prevalence of the sufficient local conditions for tropical cyclone formation and intensity change? Several recent studies have examined interannual changes in basin-scale circulations (including SST variations) and tropical cyclones, e.g. – Bell and Chelliah (2006), Goldenberg et al. (2001), Chelliah and Bell (2004), Hoyos et al. (2006), Wang and Chan (2002), Chan and Liu (2004) and others. Some of the studies have found correlations between basin-scale SST and storm number and intensities on interdecadal time scales but not on shorter time scales. Relationships between basin-scale SST variations and the conditions that cause cyclogenesis have not yet been established. Some studies find that changes in basin-scale vertical shear patterns modulate the number of tropical cyclones that occur in a basin during a season, while others dispute this. Wang and Chan (2002) found that changes in spatial patterns of storm formation in the NW Pacific during El Nino could be attributed to increased vorticity in the monsoon trough and increased subsidence in the northwest portion of the basin. 2.1.5 Summary Many questions remain. If SST changes are important to cyclogenesis, is this due to the direct, local thermodynamic effects on the incipient cyclone, or to changes in the location and intensity of convection and/or the large-scale dynamics? What kind of vertical wind-shear changes enhance or suppress genesis? Does tropical wave activity affect genesis differently as the large-scale circulation of a basin varies? What role do basin-scale circulation changes play in altering the intensity and configuration of monsoon troughs? A great deal of research is needed to answer these and similar questions, and there appear to be sufficient data available to make progress in this area. Bibliography

Bell, G. D., and M. Chelliah, 2006: Leading tropical modes associated with interannual and multidecadal fluctuations in North Atlantic hurricane activity. J. Climate, 19, 590-612. Bister, B., 2001: Effect of peripheral convection on tropical cyclone formation. J. Atmos. Sci., 58, 3463–3476. Bracken, W. E., and L. F. Bosart, 2000: The role of synoptic-scale flow during tropical cyclogenesis over the North Atlantic Ocean. Mon. Wea. Rev., 128, 353–376. Chan, J. C. L., and K. S. Liu, 2004: Global warming and western North Pacific typhoon activity from an

264

observational perspective. J. Climate, 17, 4590–4602. Chelliah, M., and G. D. Bell, 2004: Tropical multidecadal and interannual climate variability in the NCEP-NCAR reanalysis. J. Climate, 17, 1777-1803. Chen, T.-C., S.-Y. Wang, M.-C. Yen and W. A. Gallus Jr., 2004: Role of the monsoon gyre in the interannual variation of tropical cyclone formation over the western North Pacific. Weather and Forecasting, 19, 776–785. Davis, C. A., and L. F. Bosart, 2003: Baroclinically induced tropical cyclogenesis. Mon. Wea. Rev., 131, 2730–2747. DeMaria, M., J. A. Knaff and B. H. Connell, 2001: A tropical cyclone genesis parameter for the tropical Atlantic. Weather and Forecasting, 16, 219–233. Dickinson, M. and J. Molinari, 2002: Mixed Rossby–gravity waves and western Pacific tropical cyclogenesis. Part I: Synoptic evolution. J. Atmos. Sci,.. 59, 2183–2196.

Elsner, J. B., and B. H. Bossak, 2001: Bayesian analysis of U.S. hurricane climate. J. Climate, 14, 4341–4350.

Frank, W. M., and P. E. Roundy, 2006: The role of tropical waves in tropical cyclogenesis. Mon. Wea. Rev., in press. Goldenberg, S. B., C. W. Landsea, A. M. Mestas-Nuñez, and W. M. Gray, 2001: The recent increase in Atlantic hurricane activity: Causes and implications. Science, 293, 474-479. Hartmann, D. L., and E. D. Maloney, 2001: The Madden–Julian oscillation, barotropic dynamics, and North Pacific tropical cyclone formation. Part II: Stochastic barotropic modeling. J. Atmos. Sci., 58, 2559–2570. Hennon, C. C., C. Marzban and J. S. Hobgood, 2005: Improving tropical cyclogenesis statistical model forecasts through the application of a neural network classifier. Weather and Forecasting, 20, 1073–1083. Hennon, C. C., and J. S. Hobgood, 2003: Forecasting tropical cyclogenesis over the Atlantic basin using large-scale data. Mon. Wea. Rev., 131, 2927–2940. Hoyos C.D., P.A. Agudelo, P.J. Webster, and J.A. Curry, 2006: Deconvolution of the factors contributing to the increase in global hurricane intensity, Science, 312 (5770), 94-97. Inoue, M., I. C. Handoh, and G. R. Bigg, 2002: Bimodal distribution of tropical cyclogenesis in the Caribbean: Characteristics and environmental factors. J. Climate, 15, 2897–2905. Karyampudi, V. M., and H. F. Pierce, 2002: Synoptic-scale influence of the Saharan air layer on tropical cyclogenesis over the eastern Atlantic. Mon. Wea. Rev., 130, 3100–3128.

Knutson, T. R., and R. E. Tuleya, 2004: Impact of CO2-induced warming on simulated hurricane intensity and precipitation: Sensitivity to the choice of climate model and convective parameterization. J. Climate, 17, 3477–3495.

Kwok, J. H. Y., and J. C. L. Chan, 2005: The influence of uniform flow on tropical cyclone intensity

265

change. J. Atmos. Sci., 62, 3193–3212 Li, T., X. Ge, B. Wang and Y. Zhu, 2006: Tropical cyclogenesis associated with Rossby wave energy eispersion of a preexisting typhoon. Part II: Numerical simulations. J. Atmos. Sci., 63, 1390–1409 Li, T., and B. Fu, 2006: Tropical cyclogenesis associated with Rossby wave energy dispersion of a preexisting typhoon. Part I: Satellite data analyses. J. Atmos. Sci., 63, 1377–1389. Maloney, E. D., and D. L. Hartmann, 2001: The Madden–Julian oscillation, barotropic dynamics, and North Pacific tropical cyclone formation. Part I: Observations. J. Atmos. Sci., 58, 2545–2558. Mann, M. E., and K. A. Emanuel, 2006: Atlantic hurricane trends linked to climate change. EOS, 87, 233-244. McDonnell, K. A., and N. J. Holbrook, 2004: A Poisson regression model of tropical cyclogenesis for the Australian–Southwest Pacific Ocean region. Weather and Forecasting: 19, 440–455. Molinari, J. and D. Vollaro, 2000: Planetary- and synoptic-scale influences on eastern Pacific tropical cyclogenesis. Mon. Wea. Rev,. 128, 3296–3307. Molinari, J., D. Vollaro, S. Skubis and M. Dickinson, 2000: Origins and mechanisms of eastern Pacific tropical cyclogenesis: A case study. Mon. Wea. Rev., 128, 125–139. Molinari, J., D. Vollaro and K. L. Corbosiero, 2004: Tropical cyclone formation in a sheared environment: A case study. J. Atmos. Sci., 61, 2493–2509. Molinari, J., K. Canavan, and D. Vollaro, 2006: Tropical cyclogenesis within an equatorial Rossby wave packet. J. Atmos. Sci., submitted. Molinari, J., P. Dodge, D. Vollaro, K.L. Corbosiero, and F. Marks, Jr., 2006: Mesoscale aspects of the downshear reformation of a tropical cyclone. J. Atmos. Sci., 63, 341-354. Pielke Jr., R. A., C. Landsea, M. Mayfield, J. Laver and R. Pasch, 2005: Hurricanes and global warming. Bull. American Meteo. Soc., 86, 1571–1575. Pratt, A., 2005: Tropical cyclogenesis forecasting skill of the Global Forecasting System (GFS) during the 2002 and 2003 Atlantic hurricane seasons. Masters Thesis, Department of Meteorology, The Pennsylvania State University.

Roundy, P. E., and W. M. Frank, 2004: Effects of low-frequency wave interactions on intraseasonal oscillations. J. Atmos. Sci., 61, 3025–3040.

Roundy, P. E., and W. M. Frank. 2004: Applications of a multiple linear regression model to the analysis of relationships between eastward- and westward-moving intraseasonal modes. J. Atmos. Sci., 61, 3041–3048.

Roundy, P. E., and W. M. Frank, 2004: A climatology of waves in the equatorial region. J. Atmos. Sci., 61, 2105–2132.

Wang, B., and J. C. L. Chan, 2002: How strong ENSO events affect tropical storm activity over the western North Pacific. J. Climate, 15, 1643–1658.

266

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES Topic 2.2 : Internal influences on Tropical Cyclone formation Rapporteurs: Kevin J. Tory1 and Michael T. Montgomery2

1Bureau of Meteorology Research Centre, GPO Box 1289 Melbourne, Australia E-mail: k.tory@bom.gov.au Phone: +61 3 9669 4578 Fax: +61 3 9669 4660

2Department of Meteorology Naval Postgraduate School Monterey, CA 93943

& Hurricane Research Division

NOAA/AOML Miami, FL 33149

Abstract This report summarizes work completed since ITWC-V that contributes to an improved understanding of the internal influences on tropical cyclone (TC) formation. The report argues the importance of low-level vorticity enhancement during TC genesis due to convergence in convective regions, both on the individual convective element scale and on the system scale. It is argued that large-scale processes essentially drive TC genesis. These large-scale processes set up a favorable environment, and initiate the mesoscale intensification mechanisms that construct the TC-scale vortex. It is argued that these large-scale processes, and a significant portion of the mesoscale processes, are represented in contemporary global NWP models. However, the finer detail not resolved by these models is believed to be important for a more complete understanding of intrinsic upscale growth mechanisms that can occur in rotating moist convective systems and the TC genesis process in particular. This has been the subject of much research in the past four years. 2.2.1. Introduction In the four years since IWTC-V, TC formation has continued to be the least understood phase of the TC life cycle. Modelling studies have pushed rapidly forward our understanding of genesis in a variety of numerical models, and have provided plausible genesis theories to be tested. TC genesis is becoming better understood in numerical models, but we still have some way to go to discover how genesis operates in the real world. Recent observational studies are beginning to show that some aspects of the genesis process observed in numerical models do indeed operate in the real world, and provide some confidence that the models may be getting the right answers for the right reasons. In the past, lack of an observational network of sufficient spatial and temporal resolution over the open ocean has been the main reason for the lagging understanding of genesis. This is still an important issue. Doppler radars have recently provided some useful information of the finer scale details of the genesis process. There are still limitations, however, in that fixed radars are land based, and require genesis to occur near land within range of the radar (Sippel et al. 2006), and aircraft-based radar can get to the genesis area but cannot provide a continuous time record of observations (Reasor et al.

267

2005). Together, high-resolution, cloud-resolving modelling studies and the slowly improving observational network are beginning to reveal some of the mysteries of TC formation. In section 2.2.2 of this report, a historical review of genesis theories that focuses on the internal dynamics of genesis is presented that contrasts the top-down versus bottom-up debate. It is suggested that a loose genesis definition was partly responsible for the disagreement between the proponents of each theory, and that the debate has been long-lived because insufficient observations have existed to prove or disprove either theory. In section 2.2.3 it is noted that inconsistent genesis definitions exist throughout the genesis literature, and it is proposed that genesis be recognised as a process consisting of two stages (e.g., Karyampudi and Pierce; 2002): (i) the sub-synoptic scale organisation of an environment favorable for genesis (genesis preconditioning); and (ii) the mesoscale focussing of the larger-scale environmental vorticity into a TC-scale vortex. In section 2.2.4 recent modelling results are presented. Higher resolution cloud-resolving models are shown to exhibit fine-scale vortex formation and interactions very similar to those in recent observational studies. Perhaps surprisingly, coarser resolution models with convective parameterization, which do not resolve such features, have been shown to have good success in genesis forecasting. In section 2.2.5 the implications of this result is explored and the suggestion is made that TC genesis might be largely driven by scales resolvable by these coarser resolution models. 2.2.2. Top-down versus bottom-up: The debate moves forward. 2.2.2.1 Theoretical ideas revisited Two main groups, Liz Ritchie and colleagues and Michael Montgomery and colleagues have provided contrasting views of internal influences on TC genesis over the last 10 years or so. Both groups recognised the importance of Mesoscale Convective System (MCS) dynamics in driving the genesis process on the mesoscale level, and both established conceptual models of vortex enhancement leading to TC genesis.

Figure 2.2.1: Radial average of vorticity for merger of midlevel vortices in a baroclinic model with no diabatic heating. In (a) time = 72 h and (b) time = 120 h there is no background vorticity. In (c) time = 120 hours, background vorticity equivalent to three times the planetary vorticity was included (contours: 2.0 × 10−5 s−1). Images taken from Fig. 12 and 13 of Ritchie and Holland (1997). Ritchie and colleagues based their dynamical understanding of MCSs on the well-documented mid-latitude terrestrial MCS, and developed a vortex intensification theory based on vortex interactions of MCS vortices (Ritchie and Holland 1997; Simpson et al. 1997). This theory was based on observations of MCS behavior in TC genesis environments that strongly resembled vortex interactions

268

as represented in two-dimensional vorticity dynamics (Dritschel and Waugh 1992). They proposed that the MCS in a genesis environment consisted of a large stratiform precipitation region with an accompanying Mesoscale Convective Vortex (MCV), and that this mid-level vortex is responsible for the vortical behavior of the MCS cloud masses evident in visual and infrared satellite imagery. Ritchie and colleagues demonstrated with idealized models that merger of mid-level vortices in a cyclonic environment resulted in a more intense vortex with increased horizontal and vertical scale. This is illustrated in Fig. 2.2.1. They advocated a top-down hypothesis whereby a number of MCV interactions leads to a resulting vortex that eventually reaches the surface, kick-starting the hurricane heat engine. They also show that the vertical penetration is proportional to the background rotation (Fig. 2.2.1c).

Figure 2.2.2: Contours of PV versus x and y on z = 0, z = 0.25, z = 0.5, z = 0.75, and z = 1, as well as a plan view of contours of PV versus x and y on z = 0 for the midlevel vortex with single-cluster convection at (a) time t= 0, and (b) t=7 days. In (a) the MCV is evident in the domain center, and to the right there is a PV anomaly representing the effects of low-level convergence and upper-level divergence that might be expected to develop following an episode of deep cumulonimbus convection. In (b) the resulting vortex structure shows a near upright low- to mid-level cyclonic core with maximum intensity at low-levels. Images taken from Figs. 12 and 13 of Montgomery and Enagonio (1998). Montgomery and colleagues based their conceptual model on observations of Zehr (1992) that low-level vortex intensification followed bursts of intense deep convection. They showed deep convective-like Potential Vorticity (PV) anomalies embedded in a MCV led to vortex interactions that resulted in a near-symmetric vortex with strong low-level vorticity (Montgomery and Enagonio 1998; Enagonio and Montgomery 2001) on plausible development time-scales. This is evident in Fig. 2.2.2. Another hypothesis in the top-down category is what we call the “shower-head” theory by Bister and Emanuel (1997). They suggested that sustained precipitation in the stratiform cloud deck would

269

gradually saturate the relatively dry and cold layer below, from the top down while advecting cyclonic vorticity to the surface. A schematic summarizing their theory is presented in Fig. 2.2.3.

Figure 2.2.3: Conceptual model of tropical cyclogenesis from a preexisting MCS proposed by Bister and Emanuel (1997). (a) Evaporation of stratiform precipitation cools and moistens the upper part of the lower troposphere; forced subsidence leads to warming and drying of the lower part. (b) After several hours there is a cold and relatively moist anomaly in the whole lower troposphere. (c) After some recovery of the boundary layer Θe convection redevelops. Image taken from Fig. 13 of Bister and Emanuel (1997). While this theory of downward vorticity advection might, at first sight, appear to contradict Haynes and McIntyre’s (1987) statement that there can be no net transfer of absolute vertical vorticity across an isobaric surface, tilting of horizontal vorticity at the downdraft edges will act to oppose the local changes within the downdraft (see Tory et al. 2006b for details). If the tilting is sufficiently intense, anticyclonic absolute vertical vorticity will be generated in the vicinity of the updraft edges. For Bister and Emanuel’s downward advection process to impart a net cyclonic change the anticyclonic absolute vorticity must be eliminated. One way to achieve this result is through vortex interactions whereby anticyclonic absolute vorticity is expelled from the emerging cyclonic core (e.g., Montgomery and Enagonio 1998). This aspect was not discussed in Bister and Emanuel (1997). We believe this process, if it does operate, will not greatly enhance the low-level cyclonic absolute vorticity, because it will be continually weakened by low-level divergence associated with the

270

mesoscale downdraft. Unlike the Ritchie and Montgomery theories, this theory has not been pursued beyond the initial work. Although, recently the potential for a thermodynamic adjustment of the lower tropospheric state, by evaporation of rain, to favor near downdraft free convection is being reconsidered by Kerry Emanuel and colleagues (Personal Communication, K. Emanuel, 2006). Raymond et al. (1998) investigated the mean vorticity, divergence and vertical mass flux within MCSs associated with a number of developing TC’s observed during TEXMEX. They showed that often during the early stages of development the MCS kinematic structure resembled that of a system dominated by stratiform dynamics (vorticity maximised at mid-levels, mid-level convergence with divergence above and below, mean subsidence in the lower troposphere and upward motion above). As the systems intensified, the MCS kinematic nature became increasingly more convective (vorticity intensifying at lower-levels, lower tropospheric convergence, upper tropospheric divergence, mean tropospheric updrafts). They noted that the transition appeared to be coupled with an increase in the mid-level relative humidity. These kinematic observations are consistent with all three theories above. The theories differ in that the top-down theories suggest that low-level convergence becomes important after genesis is complete, whereas the bottom-up theory suggests that low-level convergence is an integral part of the genesis process. 2.2.2.2 Confusion surrounding loose genesis definition, and understanding of genesis progression

During the last decade, the debate between the two camps has been complicated by a loose definition of TC genesis, which is yet to be tied down. The top-down theories were based on the premise that at an early stage in their lifecycle MCSs typically have a MCS-scale surface anticyclone below the stratiform precipitation deck (commonly observed in mid-latitude terrestrial MCSs, e.g., Fritsch et al., 1994; Houze, 2004). Thus these theories focused on mechanisms that erode the surface anticyclone and replace it with cyclonic vorticity. In a number of earlier top-down studies (e.g., Bister and Emanuel 1997; Simpson et al. 1997; Ritchie and Holland 1997; Harr et al. 1996a,b) it has been suggested that TC genesis is the process that replaces the surface anticyclone. In these studies, the role of tropical waves in the genesis preconditioning process was recognised, as was the role of vortex enhancement in MCSs embedded in the preconditioned environment (as mentioned above). They also recognised that relatively large-scale regions of near downdraft-free convection played an important role in amplifying the TC vortex at low levels, and that low-level cyclonic vorticity must be present in the convective region before such amplification can take place. They recognised also that before near downdraft-free convection can develop the typically observed “onion-shaped” temperature and dewpoint profiles must be replaced with a moist-neutral profile. (Bister and Emanuel were not that specific. They commented only on moistening below the MCS.) Thus there appeared to be two roadblocks to TC genesis, one kinematic (low-level anticyclone) and one thermodynamic (“onion-shaped” profile). The top-down merger theory focused on the kinematic roadblock, while the top-down showerhead theory considered both roadblocks. Whether the top-down mechanism brought sufficient cyclonic vorticity to the surface to kick-start the hurricane heat engine, or generate a warm-core TC scale vortex was not of great importance to these theories. The Montgomery camp has suggested deep-convective, low-level vortex enhancement was taking place within MCSs well before a TC-scale vortex had formed, and well before the system-scale vortex became self-sustaining through a positive feedback between the surface winds and sea-to-air fluxes of latent and sensible heat from the underlying ocean. That is, sufficient low-level cyclonic vorticity was assumed to already be present in their MCS conceptual model, at least in the convective regions of the MCS. Thus, if there was ever any need to replace anticyclonic vorticity at the surface with cyclonic vorticity it must have occurred well before genesis took place. Their recent modeling studies (summarized below) suggest that vorticity enhancement in convective regions can proceed prior to the establishment of a moist-neutral lower-troposphere. At this point a specific definition of genesis might have helped clarify the debate. If genesis was said

271

to be complete once a lower-tropospheric warm-cored TC-scale vortex had formed, then the top-down proponents would likely have agreed that the genesis process must continue beyond the arrival of cyclonic vorticity to the surface and include the low-level vortex enhancement in deep convective regions (which they believed was necessary to amplify the already formed TC) to bring about the lower tropospheric, warm-core structure. In recent modelling and observational studies discussed below there does not appear to be any evidence that surface anticyclones below MCSs hamper the genesis process. It may be that warming from ocean heat fluxes weakens the surface cold layer, which in turn reduces the low-level divergence and the potential for the development of substantial anticyclonic relative vorticity, when compared with the terrestrial midlatitude MCSs (e.g., as discussed in Fritsch et al. 1994). On the other hand, significant anticyclonic relative vorticity caused by low-level divergence, becomes less likely with proximity to the equator (assuming the MCS initially developed in a low-level cyclonic absolute vorticity environment). This is because divergence cannot change the sign of absolute vorticity. It can only weaken the absolute vorticity magnitude. In which case the anticyclonic relative vorticity magnitude can at most approach the magnitude of the planetary vorticity. Furthermore, if the dynamics change and divergence is replaced with convergence (e.g., the previously stratiform area becomes dominated by deep convection) the weak anticyclonic relative vorticity will become cyclonic as the cyclonic absolute vorticity intensifies, and the kinematic roadblock has been overcome. These studies suggest that vorticity enhancement in deep convective regions with pre-existing cyclonic vorticity plays an important role from the very early stages of genesis, and while MCV merger can and probably does take place it would not appear to be necessary for genesis. 2.2.2.3 Bottom-up theory evolves The continuation of the top-down versus bottom-up debate was in part due to insufficient observational evidence to suggest with sufficient certainty that either process actually occurred in the real atmosphere. Bister and Emanuel (1997), Ritchie and Holland (1997), Simpson et al. (1997) all attempted to demonstrate their respective top-down mechanisms in observational studies, but we believe there was insufficient evidence to prove or disprove any of the theories in any of the examples. Until recently the Montgomery camp focussed on modelling to develop their theory. They began with highly idealised models in which the bottom-up theory was born, moved to high-resolution, cloud-resolving forecast models (Hendricks et al., 2004) and high-resolution cloud resolving idealized models (Montgomery et al., 2006). They found that vortical hot towers (VHT) on scales of 10—20 km played an essential role in both the realistic and idealised genesis studies. The VHT vortical structure and diabatic heating are illustrated in Fig. 2.2.4. These vortices provided seed vorticity that contributed to a vortex upscale cascade, and net heating that fuelled a system scale intensification (SSI) process, both deemed essential for genesis. (An example of vortex upscale cascade taken from Hendricks et al. (2004) is presented in Fig. 2.2.5.) The SSI process enhances vorticity on the system-scale by converging pre-existing and convectively intensified cyclonic vorticity via the Eliassen circulation within a quasi-balanced vortex driven primarily by latent heating within hot tower cores. These results were greeted with skepticism. When the VHT theory was presented VHTs were largely unheard of, which made broad acceptance of the theory more improbable. Until recently observations of VHTs on this scale have been very elusive. This is not surprising given that they are likely to be obscured by the larger scale convection in which they are embedded, and their small scales makes them difficult to identify in more traditional observational networks including aircraft flight data.

272

Figure 2.2.4: Absolute vorticity f + in units of 10−4 s−1 (left) and diabatic heating rate in units of K h−1 (right) on horizontal surfaces z = 1, 4, and 7 km at t = 7 h into the control simulation. Axes are in km. Image taken from Fig. 3 of Montgomery et al. (2006).

273

Figure 2.2.5: Plan views of absolute vertical vorticity at low (left panels) and midlevels (right panels) of the troposphere. Note the merger of the two anomalies at low-levels. Image taken from Fig. 11 of Hendricks et al. (2004). 2.2.3. Genesis definitions The genesis definition problem is not just confined to the dynamical processes of vortex enhancement on the mesoscale. The genesis process is usually identified as a synoptic-scale pre-conditioning followed by a mesoscale organisation into a TC-scale vortex (e.g., Karyampudi and Pierce, 2002). In some articles the distinction is blurred and it is not clear whether the authors are referring to a synoptic or mesoscale vorticity enhancement process. Karyampudi and Pierce (2002) describe the synoptic scale pre-conditioning as Stage 1 of genesis (e.g., recent studies include Molinari et al. 2000; Bracken and Bosart 2000; Dickinson and Molinari 2002; Li et al. 2003), and the mesoscale generation and interaction of vorticity anomalies associated with one or more MCSs as Stage 2. We will adopt this terminology in the remainder of the report. Clearly the topic of internal influences on TC formation is associated with Stage 2 of genesis, and this will remain the focus of the report. However, it should be noted that the larger-scale environment can influence the location of the convective forcing that

274

initiates Stage 2 (e.g., Chen et al. 2004), which could then influence the TC formation rate, intensity and track (e.g., Tory et al. 2006c). 2.2.4. Recent modelling results and observational studies 2.2.4.1 The collective effect of VHTs The high-resolution, cloud-resolving forecast of Hurricane Diana described in Hendricks et al. (2004) was the first study to quantitatively analyse the real genesis potential of vortex enhancement by convective-scale updrafts. Their 3 km grid spacing led to minimum convective scales of approximately 12—15 km, with associated VHTs of that scale and larger. Although their sensitivity experiments with grid spacing of 14 km produced smaller convective scales and smaller-scaled VHTs, the basic path to genesis was unchanged. The idealized study of Montgomery et al. (2006) showed that the basic genesis mechanisms of vortex upscale cascade and SSI evident in the Hendricks et al. study of Hurricane Diana was a consistent result in high-resolution, cloud-resolving models (MM5 and RAMS, respectively). Doubts about the realism of this genesis path should then be directed toward the cloud physics, initial conditions and perhaps the model resolution. A recent observational study of the formation of Hurricane Dolly (1996) by Reasor et al (2005) has provided evidence that the VHT path, as described in Hendricks et al. (2004) and Montgomery et al. (2006), does happen in the real world. Reasor et al. (2005) analysed Doppler radar data from the genesis of Hurricane Dolly to illustrate vortex enhancement in a MCS. Their analysis suggested mid-level vortex enhancement in the stratiform precipitation region, consistent with the mid-latitude terrestrial MCS theory, as well as low-level vortex enhancement on the convective scale consistent with the VHT theory. Sippel et al. (2006) also identified intense vortices associated with individual cumulonimbus in Doppler radar data of varying scales from 1.5—5 km. Although smaller in scale than the VHTs discussed in Hendricks et al. (2004) and Montgomery et al. (2006), these vortices exhibited similar vortex interactions during the formation of Tropical Storm Allison (2001). These observational studies suggest that the minimum VHT scale is about 3—4 km in diameter. Both studies show meso-β-scale (10—100 km) vortices with meso-γ-scale (1—10 km) vortices embedded, which is suggestive of stratiform and convective-type vortex enhancement at work simultaneously. Both also show a “feeder band” of convection leading to the dominant vortex core with multiple meso-γ-scale vortices embedded. This is illustrated in Fig. 2.2.6 from Reasor et al. (2005) and in Fig. 2.2.7 from Sippel et al. (2006).

4 An additional 1 km resolution study is underway as part of Marc Hidalgo’s PhD thesis, due to be finished in October,

2006.

275

Figure 2.2.6: Three-dimensional iso-surface of the vertical component of relative vorticity (15 × 10−4 s−1) between 1 and 7 km, and the horizontal winds at a height of 1 km, observed during the formation of Hurricane Dolly on 19 August 1996. Shown is the development of the low-level vortex at 2128, 2245, and 2345 UTC (clockwise from top). The feeder band about vortex V4 intensifies and spirals in toward the vortex core. Image taken from Fig. 12 of Reasor et al. (2005). With time, the smaller scale vortices merged as they were fed into the vortex core. Many of these vortices were relatively shallow. Sippel et al. (2006) found the typical radii of the vortices were less than 5 km, which was twice the size of the parent convective cell. It would be interesting to know whether a similar ratio between the convective cell size and associated vortex also applies to larger scale cells and indeed deep convective regions more generally. The vorticity estimates ranged from 10-3 to 10-2 s-1, with a peak of near 1.5×10-2 s-1. Sippel et al. (2006) commented that these scales and intensities are comparable with tornadic mesocyclones documented by Spratt et al. (1997) in TC rainbands.

276

Figure 2.2.7: The 0.3° KHGX radial velocity and reflectivity images from 5 June 2001 during the formation of Tropical Storm Allison. For reference, meso-γ-scale vortices discussed in the text are located at the center of the circles. The circle radius is approximately twice the associated vortex radius of the maximum winds. The velocity scale is in m s−1, and the reflectivity scale is in dBZ. A shear/rainband axis is denoted by the white (black) dashed line in velocity (reflectivity) data. The convection enclosed by the dotted line is a “central core” of convection. Image taken from Fig. 13 of Sippel et al. (2006). Now that VHTs of similar scale to those modeled by Hendricks et al. (2004), Montgomery et al. (2006) and Davis and Bosart (2006, see below) have been observed exhibiting similar behavior with respect to the upscale vortex cascade, the VHT route to TC genesis must be recognised as a possible route to genesis.

277

Davis and Bosart (2006) performed a high-resolution, cloud-resolving simulation of Hurricane Humberto (2001). They also found VHT-like activity played an important role in the genesis process, as in Hendricks et al. (2004) and Montgomery et al. (2006). Vorticity budgets showed the dominant vortex enhancement mechanism was low-level vorticity convergence in deep convective regions. They found significant condensational heating from convective updrafts drove “the transverse circulation necessary for the spin up of the azimuthal mean vortex” (SSI process). Humberto formed in a sheared environment, which favored convection in the downshear-left quadrant of the nacsent cyclone. Perhaps the most interesting feature of the Davis and Bosart (2006) simulation is the identification of the mesoscale convective behavior that led to the early construction of the low-level vorticity feature that became the cyclone core. Convection was initiated in a region of “warm-air advection” in the vicinity of a large-scale convergence line. Convection formed along this line by 12 hours into the simulation. The convective line had a similar structure to a mid-latitude terrestrial MCS squall line (e.g., recent studies include Bryan and Fritsch 2000; James et al. 2005, 2006), including strong line-perpendicular, low-level inflow and deep convection along the line with what appears to be a stratiform precipitation deck behind. Evaporative cooling is significant in the first 100 km behind the line, which contributes to low-level divergence. Perhaps because the stratiform rain-induced lower-tropospheric cooling was concentrated within 100 km of the convective line, the magnitude of this cooling was significant. A bore began to propagate away from the convective line (Mapes 1993), which resulted in lower tropospheric lifting and associated moistening and cooling (about 2 K) in the lower half of the troposphere. Within a 200 by 300 km region behind the bore (as well as the leading edge of the bore) deep convection developed with weakened downdrafts, which they attributed to the higher humidity. This convection is evident in Fig. 2.2.8a, which shows convective activity between two lines roughly oriented southwest-northeast, one in the lower-right of the panel (the squall line front) and the other near the center of the panel (leading edge of the propagating bore). The resulting mean convergence served to enhance the pre-existing cyclonic absolute vorticity and generate the low-level nascent cyclone within about 6 hours of the convective line forming. The enhanced low-level cyclonic relative vorticity is evident in Fig. 2.2.8b. Davis and Bosart (2006) suggested that an upper-level PV anomaly played an important role in this process by providing vertical wind shear that helped set up the convective line MCS dynamics. Note this environment was neither saturated nor moist neutral, which are conditions believed to be necessary for downdraft-free convection. Davis and Bosart (2006) commented that the cooling and moistening caused by the passage of the bore led to a reduction in downdraft intensity, which improved the efficiency of cyclonic vorticity production in the boundary layer. From this study, Hendricks et al. (2004) and Montgomery et al. (2006), it would seem that downdraft free convection may not be essential for TC genesis. Vertical wind shear also favored mesoscale regions of ascent in the developing vortex due to the dynamic response of a tilted vortex that leads to rising motion on the down-tilt side (e.g., Raymond and Jiang 1990; Raymond et al. 1998; Jones 1995). Mesoscale ascent in both the bore and on the down-tilt side of the developing TC-scale vortex led to cooling and moistening which reduced the stability of the lower troposphere and reduced the potential for downdrafts. These conditions favor the development of convective regions with net low- to mid-level convergence.

278

Figure 2.2.8: Model-derived cloud-top temperature and 10-m wind for (a) 15 UTC 19 Sept. (15 h forecast) and (b) 12 UTC 20 Sept. (36 h forecast), from a simulation of the genesis of Hurricane Humberto (2001). Image taken from Fig. 7 of Davis and Bosart (2006). The propagation of a “cold” bore behind an MCS squall-line with its associated convection, is another example of how MCS dynamics can lead to the generation of a relatively large-scale convective area. Of note is that this development occurred quite early in the Stage 2 genesis process, and that the large-scale convective area developed only about 6 hours after the initial convective line appeared. This time delay has positive implications for coarse-resolution NWP models that favor net deep convection as soon as convection is initiated (i.e., they do not resolve the finer-scale processes that lead to the ultimate convective region). If it only takes six hours for the convective region to develop, then the coarse-resolution NWP models may not be accelerating the genesis process at too great a rate. On the other hand, if the development of the convective region is sensitive to its environment as Davis and Bosart (2006) suggest, and the coarse-resolution models are not representing critical thermodynamic and kinematic processes, then false alarms may be expected. This is discussed further in the next sub-section. Davis and Bosart (2006) comment that baroclinicity (isentropic uplift) played a role in the formation of Humberto by producing a favorable region for convection, but they believe the vertical wind shear was critically important for maintaining the convection sufficiently for genesis to be successful. They hypothesized the shear acted in two ways to enhance convection. It enabled the squall line to develop (in their sensitivity experiment with weakened shear the bore that provided the moistened lower troposphere failed to develop). Shear also tilted the developing TC scale vortex, which led to mesoscale ascent on the down-tilt side. This ascent also served to moisten and cool the lower troposphere, which they suggested reduced the intensity of downdrafts. Thus, they postulate that in the deep tropics where the baroclinicity is likely to be very weak, shear could play an important convection-maintenance role in TC genesis. 2.2.4.2 GLAPS Tory et al. (2006a,b,c) have documented in detail the TC genesis process in an operational forecast model. The model has been developed from the Australian Bureau of Meteorology LAPS NWP model specifically with TC genesis in mind, and thus has recently been termed GLAPS

279

(Genesis-LAPS). The model grid spacing is 0.15° lat/long. At this resolution, convective parameterization is required. Despite the relatively coarse resolution compared with Hendricks et al. (2004), Montgomery et al. (2006), and Davis and Bosart (2006), and the use of convective parameterization instead of explicit convection, the route to genesis is surprisingly similar. Convective updraft regions of 60 km diameter and greater enhance low- to mid-level vorticity mostly through vorticity convergence, driven by the net warming by latent heating in VHT cores. The individual vortex cores that develop in the convective updraft regions interact similar to observed MCSs as they are advected around the monsoon circulation in which they are embedded (i.e., a larger scale vortex upscale cascade). An example of this behaviour taken from Tory et al. (2006b) is illustrated in Fig. 2.2.9. The PV anomaly labelled A is associated with an old convective burst undergoing decay, while B is associated with a young developing convective burst. A is a stronger PV anomaly, but as Fig. 2.2.9b shows anomaly B is intensifying rapidly, while the two anomalies rotate about each other and the monsoon low they are embedded in. Fig. 2.2.9c shows B is now considerably more intense, and A has weakened and is being drawn into B. Two hours later Fig. 2.2.9d shows that B has all but consumed A, and another anomaly C has begun to develop. With time the process is repeated with C taking over as the dominant anomaly, which then consumes B after the convective burst responsible for B decays. To help describe these processes, Tory et al. (2006a) defined primary and secondary vortex enhancement mechanisms, in which the vortex enhancement in individual updrafts (through vorticity convergence and vertical advection5) is the primary mechanism, and the vortex upscale cascade and SSI processes are secondary mechanisms. It could be argued that convection on scales resolved by GLAPS is unrealistic. However, convective regions of such scale and larger are often observed in the pre-Stage 2-genesis environment (e.g., Harr et al 1996a,b; Ritchie and Holland 1997; Simpson et al. 1997; Ritchie 2003; Ritchie et al. 2003). This is the scale of Sippel et al. (2006) meso-β vortices, or convective burst vortices (CBV). If we accept that convective regions do exist of this scale we then ask, does the modelled vortex structure resemble observed MCS features? It has been argued for some time that these structures should be viewed as largely stratiform with an accompanying MCV, in which case there would be minimal low-level cyclonic vorticity. Large areas of very cold cloud top temperatures suggestive of nearly contiguous deep convection on scales of 100 km are often observed (Zehr 1992) during the late pre-Stage 2-genesis development, with evidence of enhanced low-level vorticity. In such environments, a deep vortex core structure is more likely to accompany the cloud mass, as was present in the Davis and Bosart (2006) simulation, than a traditional continental MCV structure.

5 The net effect of vertical advection of cyclonic vorticity is offset by the net tilting which produces anticyclonic vorticity of equal magnitude on the edge of the updrafts. In this way Haynes and McIntyre’s (1987) theory is not violated (Tory et al. 2006b).

280

Figure 2.2.9: Potential Vorticity (PV, contour interval 0.5 PV units, anticyclonic hatched) on the 850 hPa surface from a G-LAPS simulation of TC Chris (2002). Vectors represent the horizontal winds on the 850 hPa surface. PV anomalies labelled A, B and C are associated with relatively short-lived convective bursts. PV anomalies outlive the convective bursts. Each frame is two hours apart. Image taken from Fig. 5 of Tory et al. (2006b). This argument is supported by observations of relatively large areas of mean low- to mid-level convergence (Zipser and Gautier1978; Mapes and Houze 1992, 1995) in a number of tropical MCSs. Additionally, observations of very low cloud-top temperatures, interpreted as overshooting deep convective cloud, on the same spatial scales and larger as the TC-LAPS updrafts are frequently observed during TC genesis (Zehr 1992; Gray 1998). However, a vortex core will only develop in this environment if it is embedded in a sufficiently cyclonic environment. Another feature often observed accompanying these large convective regions are low-level wind surges (e.g., Zehr 1992; Gray 1998). Gray has suggested that the wind surges may lead the development of the large convective blow-ups, i.e., the convergence leading the surge triggers the relatively large area of convection. Wind-surge like behavior is evident in some of the GLAPS simulations presented in Tory et al. (2006b,c), but the dynamics of the surges were not investigated. It is not clear whether the surge leads the convection or the convection leads the surge. Ritchie et al.

a

B A

b

BA

c

B A

d

BA

C

281

(2003) comment on such a surge during the development of Hurricane Floyd. The temporal resolution of their observations however was not sufficient to suggest a cause and effect relationship between the surge and updraft. The high-resolution cloud representing simulations of Hendricks et al. (2004), Montgomery et al. (2006) and Davis and Bosart (2006) are consistent with the earlier theoretical work of Montgomery and Enagonio (1998) in that the VHTs do with time interact with the vortex in which they are embedded to form a PV monolith structure similar to the GLAPS vortex core. This behavior is partially captured in the observations presented in Reasor et al. (2005). It could be argued that GLAPS forces the immediate construction of a vortex core in convective regions, rather than allowing the more gradual building process evident in the higher resolution models of Hendricks et al. (2004), Montgomery et al. (2006) and Davis and Bosart (2006). Tory et al. argue that although it is a shortcoming of the GLAPS model, it is likely to only be significant during the early Stage 2 genesis, because later observed MCS convection and low-level convergence regions are often not too dissimilar in areal extent. Furthermore, Davis and Bosart (2006) have shown that in certain situations the development of relatively large convective areas can occur in about six hours after the initial convection appears. The SSI process plays a significant role in GLAPS TC genesis also. It is likely to play a greater role in proportionate sense to the higher resolution models if, as suggested above, the finer scale of the vortex upscale cascade has been bypassed. Despite the GLAPS shortcomings identified above, the system reproduces many realistic aspects of genesis and Tory et al. (2006c) claim it has considerable success in forecasting both genesis and non-development, although an objective test of GLAPS performance is yet to be completed. The GLAPS simulations show convection typically develops about 200 km from the large-scale cyclonic circulation center, and with time as the system intensifies the convective regions and the associated vortex cores spiral inward. This process was often observed by Zehr (1992) and evidence exists of such behavior in satellite imagery provided in Harr et al. (1996b), Ritchie and Holland (1997), Simpson et al. (1997), Ritchie (2003), and Ritchie et al. (2003). Perhaps the most important finding of Tory et al. (2006c) is that TC genesis is qualitatively predictable in conventional NWP forecast models (although GLAPS is not entirely conventional in its initialization, see below), which is contrary to claims that TC genesis will forever be a largely unpredictable process (e.g., Davis and Bosart 2002). Davis and Bosart (2002) based this comment on their experience with sensitivity experiments on a particular hurricane forecast, and the comment referred to intensity and track forecasts in addition to genesis. Their expectations are likely to be more quantitative than Tory et al., who only consider the simulated genesis of a TC within a few hundred km and within about 12 hours of that observed to be a success. Tory et al. believe that qualitative TC genesis forecasting is predictable because the genesis process appears to be driven by large-scale dynamics. Provided the initialization is able to capture the large-scale environment, the model should be able to capture the genesis process, because the large-scale environment satisfies Gray’s necessary genesis conditions and it contains the dynamical forcing that initiates convection. If the coarser resolution models such as GLAPS adequately capture the net effect of the unresolved vortex interactions then these coarser resolution models should also have qualitative success in genesis forecasting. A critical component to GLAPS is the initialization, which implants artificial heat sources in locations where very low cloud-top temperatures are observed. This ensures convection is active at the initial time in the correct location, and does not rely on the model to determine where convection first appears. Davis and Bosart (2002) commented that for the most accurate forecast various physical parameterizations would need to be tuned, and that such tuning is not likely to be consistent from event to event. In GLAPS, the arbitrary heat sources have been tuned to this effect, but broad success has been achieved by tuning to a number of borderline TC genesis cases, rather than just one case. Tory et al. (2006c) agree with Davis and Bosart that the finer detail of the evolution can be quite sensitive to

282

initial conditions, but overwhelmingly they found that the final result in terms of vortex size, intensity and location, differed very little. This result led Tory et al. (2006c) to hypothesize that on a qualitative level the finer details of vortex numbers and sizes, and the details of vortex interactions were relatively unimportant to the overall genesis result. Instead, what was important was the state of the larger scale environment in which the system develops and the net convection that took place. They base this on the apparent importance of the SSI process, which is essentially a function of the large-scale cyclonic environment and the net heating of all convective elements. 2.2.5. Some implications of these results Forecasters in the Australian region (and perhaps in other areas of the world) regularly use global NWP models for TC genesis guidance. The relatively coarse resolution of these models supports the argument above that genesis is on the whole predictable and that it is driven by large-scale processes. This conclusion was also reached by Camargo and Sobel (2004) in their study of TC genesis in low-resolution atmospheric GCM. Australian forecasters use ECMWF global forecasts (Jeff Callaghan, personal communication, 2006). As a rule of thumb, once a single closed isobar has formed they consider genesis to be complete. Furthermore, methods for determining tropical cyclone-like structures in NOGAPS have been developed, not so much based on recognized TC structure but on the structures present in the global model when TCs are observed (Cheung and Elsberry 2002). It could be argued that at such coarse resolution (in which the finer details cannot be resolved) the models must be getting the right answer for the wrong reason. But if they consistently get the right answer for this supposedly “wrong” reason then we must question the importance of the finer detail of TC formation. The pioneering work of Kurihara and Tuleya (1981) showed that coarse resolution (0.625° lat/long grid spacing) NWP models are capable of simulating the development of a realistic tropical storm. They commented that the warm core resulted from excess heating produced by the diabatic heating- adiabatic cooling imbalance, which essentially describes the SSI process. At such resolution, there was little evidence of a vortex upscale cascade, although an elongated vorticity anomaly became increasingly symmetric as it formed the core of the TS, and a remote anomaly appeared to orbit and be sheared by the TS vortex. In this coarse resolution case, the dominant vortex enhancement mechanisms (using Tory et al.’s terms) are the primary mechanism of vortex stretching in the updraft regions, and the secondary mechanism of SSI. The role of the vortex upscale cascade would be to simply focus the widely distributed anomalous cyclonic vorticity into a central core while expelling anomalous anticyclonic vorticity (Montgomery et al. 2006; Tory et al. 2006a,b). It is conceivable that if the net vorticity and net heating between two simulations, in which one has high resolution and captures the intricate detail of vortex upscale cascade, and one has low resolution in which the vortex upscale cascade is reduced to the simple axisymmetrization of large-scale anomalies within the central vortex, then the resulting systems could have very similar intensity. In the same comparison, the importance of capturing the details of the vortex upscale cascade is likely to be much greater for accurate forecasts of genesis and intensification rates, and track accuracy. 2.2.6. Conclusions To be consistent with the theory of Haynes and McIntyre (1987), which states that there can be no net transfer of absolute vorticity (PV) across a pressure (isentropic) surface, TC genesis must consist of a series of processes that cause a near-horizontal redistribution of absolute vorticity into an upright vorticity (PV) monolith. The process begins with the large-scale redistribution of absolute vorticity into cyclonic and anticyclonic circulations by tropical waves. The cyclonic components may often be amplified by convergence within the developing wave field. This large-scale vorticity redistribution, which can be considered Stage 1 of genesis, provides dynamical pre-conditioning for Stage 2. Stage 2 of genesis is driven by MCSs. The thermodynamic conditions necessary for Stage 2 genesis are nearly always met in the summertime, tropical oceanic environment where genesis is common, and for that reason the study of Stage 2 TC genesis has focussed mostly on the dynamics of vortex

283

enhancement in MCSs. These MCSs may develop within the pre-conditioned environment where weak baroclinicity favors convection (mesoscale ascent along sloping isentropic surfaces). We suggest that the processes that favor convection are resolvable in coarse resolution NWP models. Contemporary global NWP models (e.g. ECMWF medium-range forecast model) have some success representing TC genesis. Although these models cannot reproduce the finer details of the vortex upscale cascade, the SSI process is represented. The relative importance of the SSI and vortex upscale cascade processes in constructing the TC-scale vortex monolith is not known, but the level of success of the global NWP models suggests the SSI process is important. The modeled convective regions consist of net low- to mid-tropospheric convergence, which enhances the low- to mid-tropospheric vorticity, and ultimately leads to the formation of a TC-scale vortex monolith. The process by which this occurs is complex and varied, but ultimately follows the same basic rules. Deep convective regions converge low-to mid-tropospheric absolute vorticity on the updraft scale and advects absolute vorticity upwards, which deepens the vortex cores (the primary enhancement mechanism). Anticyclonic vorticity forms on the edge of these updrafts, consistent with the constraints of Haynes and McIntyre (1987). The cyclonic cores interact to form larger cores and to expel anticyclonic vorticity (the vortex upscale cascade). The diabatic heating slightly outweighs the adiabatic cooling in these cores, and the net effect of multiple warm vortex cores is the amplification of the system scale secondary circulation (the SSI mechanism). The modeling and observational results presented above on the whole contradict the top-down theories, which focused on mechanisms to bring mid-level cyclonic vorticity down to the surface. While it is recognised that convergence into stratiform precipitation regions can generate significant mid-tropospheric vortices and that merger of such vortices can lead to larger and deeper vortices, the modeling studies suggest it is the deep convective vortex enhancement mechanisms that are responsible for the development of a warm-cored, TC-scale vortex. It would seem likely that mid-tropospheric vortex enhancement in the stratiform precipitation region, and merger of these vortices is likely to increase the probability of genesis, by enhancing the MCS cyclonic environment. 2.2.6.1 Some remaining questions about the physics of TC formation

The following questions address the uncertainty that surrounds the thermodynamic roadblock mentioned in Section 2.2.2.2. How does the MCS environment become transformed from the “onion-shaped” temperature and dewpoint profiles, to a moist neutral profile? (The assumption here is that the deep convective dynamics, that we know are essential for enhancing the low-level vorticity during TC formation, cannot proceed until we have downdraft-free convection on the MCS-scale.) The VHT route to genesis suggests that vortex enhancement can proceed without this adjustment to the thermodynamic profile on the MCS scale. This result motivates the following questions: Is downdraft-free convection on the MCS scale essential for TC genesis? Could the VHT activity generate a moist neutral lower troposphere necessary for downdraft-free convection on the MCS scale? What conditions are necessary for VHT generation, and what processes lead to the formation of such conditions? In order to enhance the lower-tropospheric vorticity on the MCS scale, how far can we deviate from moist neutrality and downdraft-free convection? Are convective regions that have diameters of 100 km, observed by Zehr (1992) and others in TC genesis environments, made up of downdraft-free convection?

284

What processes lead to the formation of these conditions on such scales? If downdraft-free convection is necessary for TC genesis on the MCS scale, and if a moist neutral lower troposphere is necessary for downdraft-free convection, how can development proceed in a weak to moderate sheared environment, in which dry air is entrained into the MCS from outside? 2.2.7 Recommendations 1. Establish a widely accepted genesis definition. Suggestion: Treat genesis as a two-step process that concludes with the establishment of a warm-cored TC-like vortex. Consider the establishment of sub-synoptic environment favorable for genesis to be Stage 1. Consider the mesoscale organisation of this environment into the warm-cored TC-like vortex to be Stage. 2. Continue investigating the thermodynamic evolution of MCSs in the tropical oceanic environment. 3. Continue investigating the nature of vorticity enhancement in MCSs that develop in the genesis environment. 4. Determine why most tropical disturbances fail to become warm-cored, surface- concentrated vortices. Acknowledgements We’d like to thank Jeff Kepert, Noel Davidson, Ed Zipser, and Jeff Callaghan for many informative discussions, and we greatly appreciate Russ Elsberry’s constructive comments on this report.

Bibliography Bister, M., and K. A. Emanuel, 1997: The genesis of Hurricane Guillermo: TEXMEX analyses and a modeling study. Mon. Wea. Rev., 125, 2662—2682. Bracken, W. E. and L. F. Bosart. 2000: The role of synoptic-scale flow during Tropical Cyclogenesis over the North Atlantic Ocean. Mon. Wea. Rev., 128, 353–376. Bryan, G. H., and M. J. Fritsch. 2000: Moist absolute instability: The sixth static stability state. Bull. Amer. Met. Soc., 81, 1207–1230. Camargo, S. J. and A. H. Sobel, 2004. Formation of tropical storms in an atmospheric general circulation model, Tellus: 56A, 56-67. Chen, Tsing-Chang, S.-Y. Wang, M.-C. Yen and W. A. Gallus Jr.. 2004: Role of the monsoon gyre in the interannual variation of Tropical Cyclone formation over the Western North Pacific. Weather and Forecasting, 19, 776-785. Cheung, K. K. W., and R. L. Elsberry, 2002: Tropical cyclone formations over the western north Pacific in the Navy operational global atmospheric prediction system forecasts. Weather and Forecasting. 17, 800-820. Davis, C., and L. F. Bosart, 2002: Numerical simulations of the genesis of hurricane Diana (1984). Part

285

II: Sensitivity of track and intensity prediction. Mon. Wea. Rev., 130, 1100-1124. Davis, C. A. and L. F. Bosart, 2006: The formation of Hurricane Humberto (2001): The importance of extra-tropical precursors. Quart. J. Roy. Meteor. Soc. (coming soon). Dickinson, M.J., and J. Molinari, 2002: Mixed Rossby-gravity waves and western Pacific tropical cyclogenesis. Part I: Synoptic evolution. J. Atmos. Sci., 59, 2183-2196. Dritschel, D. G., and D. W. Waugh, 1992: Quantification of the inelastic interaction of unequal vortices in two-dimensional vortex dynamics. Phys. Fluids A, 4, 1737. Enagonio, J. and M. T. Montgomery, 2001: Tropical cyclogenesis via convectively forced vortex Rossby waves in a shallow water primitive equation model. J. Atmos. Sci., 58, 685—706. Fritsch, J. M., J. D. Murphy and J. S. Kain, 1994: Warm core vortex amplification over land. J. Atmos. Sci., 51, 1780—1807. Gray, W. M., 1998: The formation of Tropical Cyclones. Meteor. Atmos. Phys. 67, 37—69. Harr, P. A., M. S. Kalafsky and R. L. Elsberry, 1996a: Environmental conditions prior to formation of a midget tropical cyclone during TCM-93. Mon. Wea. Rev. 124, 1693—1710. Harr, P. A., R. L. Elsberry and J. C. L. Chan, 1996b: Transformation of a large monsoon depression to a tropical storm during TCM-93. Mon. Wea. Rev. 124, 2625—2643. Haynes, P. H. and M. E. McIntyre, 1987: On the evolution of vorticity and potential vorticity in the presence of diabatic heating and frictional or other forces. J. Atmos. Sci., 44, 828—841. Hendricks, Eric A., M. T. Montgomery and C. A. Davis. 2004: The Role of “Vortical” Hot Towers in the Formation of Tropical Cyclone Diana (1984). J. Atmos. Sci., 61, 1209-1232. Houze, R. A., 2004: Mesoscale Convective Systems. Reviews of Geophysics, 42, RG4003, 1—43. James, R. P., J. M. Fritsch and P. M. Markowski. 2005: Environmental Distinctions between Cellular and Slabular Convective Lines. Mon. Wea. Rev., 133, 2669–2691. James R. P., P. M. Markowski and J. M. Fritsch. 2006: Bow Echo Sensitivity to Ambient Moisture and Cold Pool Strength. Mon. Wea. Rev., 134, 950–964. Jones, S. C., 1995: The evolution of vortices in vertical shear. I: Initially barotropic vortices. Quart. J. Roy. Meteor. Soc., 121, 821—851. Karyampudi, V. M., and H. F. Pierce, 2002: Synoptic-scale influence of the Saharan air layer on tropical cyclogenesis over the eastern Atlantic. Mon. Wea. Rev., 130, 3100-3128. Kurihara, Y., and R. E. Tuleya, 1981: A numerical simulation study on the genesis of a tropical storm. Mon. Wea. Rev., 109, 1629—1653. Li, T., B. Fu, X. Ge, B. Wang, and M. Peng. 2003: Satellite data analysis and numerical simulation of tropical cyclone formation, Geophys. Res. Lett. 30, 2122. Mapes, B. E. and R. A. Houze, 1992: An integrated view of the 1987 Australian monsoon and its mesoscale convective systems. I: Horizontal structure. Quart. J. Roy. Meteor. Soc., 118, 927—963.

286

Mapes, B. E., 1993: Gregarious tropical convection, J. Atmos. Sci., 50, 2026—2037. Mapes, B. E. and R. A. Houze, 1995: Diabatic divergent profiles in Western Pacific Mesoscale Convective Systems. J. Atmos. Sci., 52, 1807—1828. Molinari, J. D. Vollaro, S. Skubis and M. Dickinson. 2000: Origins and mechanisms of Eastern Pacific Tropical Cyclogenesis: A case study. Mon. Wea. Rev., 128, 125–139. Montgomery, M. T. and J. Enagonio, 1998: Tropical cyclogenesis via convectively forced vortex Rossby waves in a three-dimensional quasigeostrophic model. J. Atmos. Sci., 55, 3176—3207. Montgomery, M. T., M. E. Nicholls, T. A. Cram and A. Saunders, 2006: A “vortical” hot tower route to tropical cyclogenesis. J. Atmos. Sci., 63, 355–386. Raymond, D. J., C. L′opez-Carillo, and L. L′opez-Cavazos, 1998: Case-studies of developing east Pacific easterly waves. Quart. J. Roy. Meteor. Soc., 124, 2005—2034. Raymond, D. J. and H. Jiang, 1990: A theory for long-lived convective systems, J. Atmos. Sci., 47, 3067—3077. Reasor, P. D., M. T. Montgomery and L. F. Bosart. 2005: Mesoscale observations of the genesis of Hurricane Dolly (1996). J. Atmos. Sci., 62, 3151–3171. Ritchie, E. A. and G. J. Holland, 1997: Scale interactions during the formation of Typhoon Irving. Mon. Wea. Rev., 125, 1377—1396. Ritchie, E. A., J. Simpson, W. T. Liu, J. Halverson, C. S. Velden, K. F. Brueske and H. Pierce, 2003: Present day satellite Technology for hurricane research. Chapter 12, Hurricane: Coping with disaster. Ed. R. Simpson, American Geophysical Union, 360 pp. Ritchie, E. A., 2003: Some aspects of mid-level vortex interaction in tropical cyclogenesis. Chapter 12, Cloud systems, hurricanes and the Tropical Rainfall Measuring Mission (TRMM): A tribute to Dr. Joanne Simpson. Ed. W-K Tau and R. Adler. American Meteorological Society. Simpson, J., E. A. Ritchie, G. J. Holland, J. Halverson and S. Stewart, 1997: Mesoscale interactions in Tropical Cyclone Genesis. Mon. Wea. Rev., 125, 2643—2661. Sippel, J.A., J. W. Nielsen-Gammon and S. E. Allen. 2006: The Multiple-Vortex Nature of Tropical Cyclogenesis. Mon. Wea. Rev., 134, 1796–1814. Spratt, S. M., D. W. Sharp, P. Welsh, A. Sandrik, F. Alsheimer and C. Paxton. 1997: A WSR-88D assessment of Tropical Cyclone outer rainband tornadoes. Weather and Forecasting, 12, 479–501. Tory, K. J., M. T. Montgomery and N. E. Davidson, 2006a: Prediction and diagnosis of Tropical Cyclone formation in an NWP system. Part I: The critical role of vortex enhancement in deep convection. J. Atmos. Sci. (Accepted.) Tory, K. J., M. T. Montgomery, N. E. Davidson and J. D. Kepert, 2006b: Prediction and diagnosis of Tropical Cyclone formation in an NWP system. Part II: A detailed diagnosis of tropical cyclone Chris formation. J. Atmos. Sci. (Accepted.) Tory, K. J., M. T. Montgomery and N. E. Davidson, 2006c: Prediction and diagnosis of Tropical Cyclone formation in an NWP system. Part III: Developing and non-developing storms. Submitted to J. Atmos. Sci.

287

Zehr, R., 1992: Tropical cyclogenesis in the western North Pacific. NOAA Tech. Rep. NESDIS 61, 181 pp. Zipser, E. J. and C. Gautier, 1978: Mesoscale events within a GATE tropical depression. Mon. Wea. Rev., 106, 789—805.

288

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES Topic 2.3 : Operational Forecasting of Tropical Cyclone Formation Rapporteur: Jeff Callaghan Bureau of Meteorology, 295 Ann Street Brisbane 4001, Australia. E-mail J.Callaghan@bom.gov.au Fax: 61 7 32214895 Working Group: Peter Otto, Steve Ready, Todd Smith, Alipate Waqaicelua. Abstract: Tropical cyclone formation occurs mostly in data sparse oceanic regions and heavy reliance is made on satellite data to locate, analyse and forecast these processes. In the Australian region formation can occur near the coast and near observational data. We use this data to show that numerical models can at times accurately forecast tropical cyclone formation and in doing so also forecast accurately the thermal structure in the lower region of the troposphere. We also address the cases of rapid intensification of tropical cyclones in strong vertical wind shear that are often not forecast by the numerical models. 2.3.1 Introduction. Tropical cyclone formation at times is forecast extremely well by the numerical modes and we present several examples of these cases. In some other situations the models provide a general and useful hint at impending tropical cyclone formation by forecasting the development of a low pressure area in the general region of interest so that then continued monitoring is essential. At the other end of the forecasting scale there are events where the models do not forecast the rapid formation of small tropical lows into tropical cyclones and these continue to cause loss of life. We describe the mechanisms associated with this type of cyclogenesis, which was first cited as a serious problem for Island countries in the Southwest Pacific at a Coral Sea Region Tropical Cyclone Coordination meeting on 4-5 November 1999 in Brisbane. 2.3.2. Thermal structure and isentropic ascent. Many reports of torrential rainfall in the northeast Australian region indicate that this occurs mostly when the wind direction backs with height. Consistent with this are the long-term mean (more than fifty years) vertical wind profile for heavy rain events in Cairns Meteorological Station. Cairns is located at Latitude 160 52’25” South on the northeast coast of Australia. The mean wind producing 24 hour rainfall totals of 150mm or more and averaged over the 24 hours leading up to when the rain gauge is read is: -130/13 knots at 950 hPa backing to 095/17 knots at 850 hPa backing to 074/13 knots at 700 hPa and backing to 021/05 knots at 500 hPa.

Heavy tropical rain reports have been documented in Bonell, Callaghan and Connor (2005) and Callaghan and Bonell (2005). For weaker rainfall, Connor and Bonell (1998) found that warm air advection, particularly in the layer between 950 and 800 hPa, contributed significantly to trade wind precipitation amounts along the north Queensland coast. There are several studies, which focus on this

289

process of shear causing isentropic ascent, which leads to the generation of convection. These studies are Fritsch et al (1994), Raymond and Jiang (1990) and Raymond (1992). The warm and cold air advection patterns that we present here resemble in their simplest form Figure 6 in Raymond and Jiang (1990), which illustrated a 700 hPa surface where the upward (downward) motion was correlated with warm (cold) air advection. In our studies we have concentrated on this 700hPa level in an attempt to locate general areas of isentropic ascent and descent. The 850 to 500hPa vertical wind shear and thicknesses represent the thermal patterns on these 700hPa surfaces. These thermal structures can also be deduced operationally by attention to the synoptic 850 hPa and 500 hPa charts. In the simplest cases, patterns that bring cooler 700hPa air near to the cyclone, are associated with a trough system increasing in intensity with height between 850 hPa and 500 hPa or ridging decreasing with height. Therefore for an intensifying cyclone moving eastwards we see a short wave 500 hPa trough move closer to the cyclone. For a westward moving intensifying cyclone the synoptic charts show a weakness develop in the 500hPa ridge south of the cyclone. 2.3.3 Rapid cyclogenesis. Research by Holliday and Thompson (1979) indicates that 75% of all western North Pacific tropical cyclones deeper than 920 hPa have experienced a period of rapid intensification of 42hPa per day or more. Extreme deepening rates of nearly 100hPa per day have been observed. All tropical cyclones, even the weaker ones, should therefore be regarded as potentially serious. As an illustration of how quickly cyclogenesis can proceed with small tropical cyclones Figure 2.3.1 shows a weak tropical low surrounded by winds of only 10 to 15 knots at 0500UTC 26 January 1996 intensified into a category 3 tropical cyclone (Celeste) in 24 hours.

Figure 2.3.1 (left). The rapid cyclogenesis of tropical cyclone Celeste near from Townsville and Mackay radars and GMS satellite imagery bottom right panels. Figure 2.3.2 (right). The Madden Julian Oscillation (MJO) from 6 January 2006 to 12 February 2006. Examples in the Australian region include Chloe (1994 in WA) deepened from a central pressure of 995 hPa to 955hPa in 24 hours. Similarly Barry went from 990 hPa to 950hPa in 24 hours in the Gulf of Carpentaria. Coral Sea cyclones Celeste deepened from 995 to 965 hPa in 12 hours, Rona from 995 to

290

970hPa in 13 hours, and Steve from 996hPa to 980hPa in 12 hours. Recent Atlantic examples of extreme intensification rates include Wilma deepening 83hPa in 12 hours and Rita deepening 44hPa in 12 hours. 2.3.4 Predictable tropical cyclone formation associated with a burst in the Madden Julian Oscillation (MJO). In Figure 2.3.2 an active phase of the MJO can be seen to cross the Australian continent and move into the southwest Pacific from the Outgoing Longwave Radiation (OLR) sequence. This MJO helped form Western Australian (WA) tropical cyclones Clare (8 January) and Daryl (19 January), the Central Australian cyclone or the so called Landphoon (27 January), Coral Sea Cyclone Jim (28 January) and Fiji cyclone Vaianu (12 February. Below we compare the 144-hour European Centre for Medium Range Weather Forecasting Centre (EC) forecasts with the verifying analyses. The EC forecast at 0000UTC 13 January 2006 showed a 990hPa mean sea level (MSL) low just north of Broome on the northern WA coast at 0000UTC 19 January. At 0000UTC 19 January Tropical cyclone Daryl, category 1 was 100km north of the forecast position. Daryl was named 12 hours earlier. Below in Figure 2.3.3 the 700hPa forecasts and verifying analysis show similar thermal patterns at 700hPa with the main warm air advection region west of the centre where the microwave data (centre panels Figure 3) show the deep convection to be located.

291

Figure 2.3.3(top)…EC 700hPa winds (black) 850/500hPa shears (red) and 850/500hPa contours (blue) with actual observations plotted larger. Large red (blue) arrows highlight areas of warm (cold) air advection. For 144 hour forecast and verifying analyses at 0000UTC 19 January 2006. (Centre panels) 91 GHz horizontally polarised microwave image for 1220 UTC 18 January 2006 (left) and 1207UTC 19 January 2006 (right). (Lower panels) EC 200hPa winds for 144 hour forecast and verifying analyses at 0000UTC 19 January 2006 with actual observations plotted larger.

292

The upper patterns (lower panels) are also similar with a weak trough to the southeast of the cyclone and accelerating easterly flow to the west of the centre over the convection. Therefore as the large scale 850 and 500hPa flow generates the thermal structure the very accurate forecast of formation of the tropical cyclone was associated with an accurate forecast of the large scale environment of the tropical cyclone. Similarly both the EC 120 hour forecasts at 1200UTC 5 February 2006 and 1200UTC 6 February forecast a 1000hPa low to form east of Fiji. Again in Figure 2.3.4, the 120-hour forecasts and analyses at 0000UTC 11 February are compared. These are again found to be similar with the 700hPa warm air advection chiefly in the northwest quadrant where the microwave data located the deep convection (not shown) and the centre of the low being located under 200hPa diffluent flow between easterly and southerly winds. Tropical cyclone Vaianu formed in this area at 0000UTC 12 February 2006.

Figure 2.3.4 EC 700hPa and 200hPa as in Figure 3 except for 120 hour forecast and verifying analyses at 0000UTC 11 February 2006.

293

2.3.5 Genesis associated with Equatorial Rossby (ER) wave. In Figure 2.3.5 the OLR pattern shows an ER wave form near the date line and move west towards Papua New Guinea (PNG) between 9 April and 17 April. The models do not forecast genesis of intensification of tropical cyclones in the Coral Sea very well. Leading up to this event the models (especially the EC) persistently forecast the formation of a low with a closed isobar. In this region this is an indication that a tropical cyclone is very likely to form. Tropical cyclone Monica was named at 0000UTC 17 April 2006 and in Figure 2.3.6 warm air advection is seen to develop on the northern side of the circulation during the cyclogenesis period between 0000UTC 16 April and 17 April. The microwave data at 2139UTC (lower right frame) is close to being co-located with the warm air advection zone.

Figure 2.3.5 (right). Equatorial Rossby Wave from 9 April January 2006 to 17 April 2006.

294

Figure 2.3.6 700hPa analyses (left panels) as in Figure 3 except during the genesis of Monica. Microwave (right panels) depiction of the genesis of Monica near the southeast tip of PNG. 2.3.6 Genesis associated with southern Hemisphere influence. The initial vortex associated with the genesis of tropical cyclone Larry appeared to originate at middle levels from a trough system extending into the tropics from the south. This occurred during a suppressed phase of the MJO. Tropical cyclone Larry was named at 1800UTC 17 March and the 700hPa sequence (Figure 2.3.7) through the genesis period show an extensive area of warm air advection develop in the southeast quadrant with very little subsiding cold air advection near the system. The microwave sequence in Figure 2.3.8 covers the genesis period and the ultimate formation of an eye. Note how the convection is mostly confined to the southeast quadrant before wrapping around to form an eye.

295

Figure 2.3.7 700hPa analyses as in Figure 3 except during the genesis of Larry.

Figure 2.3.8 Microwave (lower panels) depiction of the genesis of Larry. 2.3.7 Tropical cyclone formation in strong vertical wind shear. This group of cyclones tend to be not properly warned for. This occurs due to their rapid rate of formation and the Dvorak (1984) satellite intensity analyses scheme does not indicate patterns which would be associated with conventional tropical cyclones. Tropical cyclone Kelvin. For the example in Figure 2.3.9, the 700hPa and 200hPa charts are at a time during the rapid genesis and intensification of tropical cyclone Kelvin when it developed from a weak tropical low into a hurricane force cyclone in 24 hours. At 700hPa there was an indication of a strong dipole of cold and warm air advection by the strong 850 to 500hPa vertical wind shears immediately

296

south of the developing cyclone. The NCEP/NCAR reanalyses showed a tropopause undulation near Mt Isa (where the temperature of -49oC is plotted. The system was moving towards 120 degrees at 18 knots and storm relative wind plots and streamlines are also shown in Figure 2.3.9. These show outflow from the developing cyclone rather than inflow from the warm tropopause undulation. Nearing peak intensity (Figure 2.3.10) it had yet to be named as the Dvorak analyses at the time indicated it was a shear pattern just below tropical cyclone intensity. However at 1230UTC 25 February 1991 (at the time of the satellite image in Figure 2.3.10) the wind at Willis Island was southerly averaging 65 knots and gusting to 85 knot. This placed the centre just outside the area of deep convective cloud tops.

Figure 2.3.9(left) 700hPa observations (as in Figure 2.3.3) except during rapid genesis of Kelvin 200hPa observations (right) and storm relative winds and streamlines.

Figure 2.3.10.. Enhanced Infrared satellite imagery of tropical cyclone Kelvin at Peak intensity near Willis Island at 12300UTC 25 February 1991. Port Moresby Cyclone 2005. The 700hPa sequence (Figure 2.3.11) shows the period of rapid intensification as the cyclone moved towards Port Moresby and causing severe damage at that centre soon after. Note how the thermal gradient and shears increase near the cyclone as a thermal trough extends up onto Cape York Peninsula. The 200hPa pattern initially (at 0000UTC 14 April 2005) was

297

diffluent over the region between Cape York Peninsula to the southeast area of PNG and this was a pattern we often see during cyclogenesis. However at 0000UTC 15 April the winds over the system were not visibly diffluent and the cyclone began rapidly weakening over the following 6 hours as it moved eastwards towards the southeast tip of PNG.

Figure 2.3.11.. 700hPa and 200hPa analyses as in Figure 3 except during the rapid genesis of the Port Moresby cyclone.

Figure 2.3.12. Rapid intensification of the Fiji 2004 cyclone from microwave data.

298

Figure 2.3.13. Enhanced Infrared satellite imagery shows rapid increase in cold cloud over a period of 5 hours.

Figure 2.3.14.700hPa and 200hPa analyses from actual observations and NCEP/NCAR reanalyses data during the rapid genesis of the Fiji (2004) tropical cyclone. Fiji Cyclone 2004. This was a small system and it proved quite difficult to track until it moved within the range of the Nadi radar. Minimum pressure recorded was 990hPa and one of the stations in Fiji reported sustained winds of 43 knots before it stopped transmitting. Numerical forecast models failed to

299

forecast the development of this system. The microwave sequence (Figure 2.3.12) clearly shows the rapid intensification of this system northwest of Nadi while the EIR sequence (Figure 2.3.13) indicated the rapid increase in convection up to 1125UTC when it would needed to have been named to provide any useful warning. The 700hPa sequence (Figure 2.3.14) has been reconstructed from the sparse actual data and a re adjustment of the NCEP/NCAR reanalyses which had the low level vortex in the wrong area. The intensification is associated with a thermal trough moving up towards Fiji. The 200hPa winds show a diffluent pattern over Fiji during the intensification. Hurricane Gabrielle. During the period covered by Figure 2.3.15 aircraft reconnaissance calculated the central pressure of Gabrielle fell from 992hPa to 972 hPa in 3 hours. As with the three southern Hemisphere cyclones above, the rapid formation and intensification occurred in the region of strong 700hPa thermal gradient. However the intensification of Gabrielle also occurred as it moved under a warm air advection zone at 200hPa. From Hirschberg and Fritsch 1993, this would force pressure falls in the centre of the hurricane. We will present examples where the interaction of a tropical cyclone and tropopause undulation has similarly been associated with rapid cyclogenesis including the so called hybrid systems.

Figure 2.3.15. 700hPa and 200hPa analyses using conventional observations, reconnaissance data and NCEP/NCAR reanalyses data. 700hPa analyses are as above but 200hPa analyses include isotherms that identify tropopause undulation from temperature maximum. The red arrows in the 200hPa analyses highlight the areas of warm air advection originating from the tropopause undulation.

300

2.3.8 Recent low latitude cyclogenesis in Appendix. In the Appendix we illustrate the rapid formation of typhoon Ewiniar. This shows the warm/cold air advection patterns even at low latitudes and the microwave indicating the convection being located on the warm air advection side of the storm. The typhoon was still a weak depression at 1200UTC 29 June 2006. At this time the EC correctly forecast the 700hPa thermal structure and advection patterns 72 hours in advance, when the rapid intensification to a super typhoon began. The favorable upper outflow structure both actual and forecast will also be presented. 2.3.9 Summary. We show how numerical models at times can provide excellent forecasts of the formation of tropical cyclones. Associated with these correct forecasts are similarly accurate forecasts of the thermal structure of tropical cyclones indicating that these structures are important to the intensification mechanism. We also discuss other events where knowledge of the model performance in the basin can alert forecasters to increase the monitoring of likely cyclogenesis. Events are also analysed where the models perform badly. Our aim here is to make forecasters aware of the patterns in which these systems evolve so as to give them the confidence to issue early warnings. We will also present null cases when strong upper winds aloft prevent or delay the cyclogenesis and also cases where low 850/200hPa vertical wind appears to favor tropical cyclone formation but the unfavorable shear is hidden in the layers between 850hPa and 500hPa.

301

2.3.10 Appendix

Figure A1 700hPa pattern as in Figure 3, showing the warm air advection pattern developing as typhoon Ewiniar formed between latitude 6 and 10 degrees North.

302

Figure A2. Formation of a microwave eye, note the main rainband on the warm air advection side of the storm and the lack of convection on the cold air advection side of the storm.

Figure A3. 700hPa pattern at the commencement of the period of rapid intensification to a super typhoon and compared with the forecast for this time from 72 hours earlier when the system was still a weak tropical depression.

303

2.3.11 Bibliography Bracken, W.E. and L. F. Bosart. 2000. The Role of Synoptic-Scale Flow during Tropical Cyclogenesis over the North Atlantic Ocean. Mon. Wea Rev. 128. pp 353–376.

Callaghan, J. J. 2004: Tropical Cyclone Intensification, Preprints, The International Conference on Storms Brisbane Queensland Australia July 2004.

Callaghan, J., and M.Bonell 2005: An overview of the Meteorology and climatology of the humid tropics. Forests, Water and People in the Humid Tropics International Hydrological Series Edited by M. Bonell and L. A. Bruijnzeel. Published by Cambridge University Press ISBN 0 521 82953 4, 925 pages

Connor, G. J. and Bonell, M. 1998. Air mass and dynamic parameters affecting trade wind precipitation on the northeast Queensland tropical coast. Int. J. Climatol., 18, 1357-1372.

DeMaria, M., J. A. Knaff, and B. H. Connell, 2001: A Tropical Cyclone Genesis Parameter for the Tropical Atlantic. _Wea. Forecasting, 16, 219-233. Dvorak, V.F. 1984: Tropical Cyclone Intensity Analysis Using Satellite Data. NOAA Technical Report NESDIS 11. 45pp

Dunion, J. P., and C. S. Velden, 2004: The impact of the Saharan air layer on Atlantic tropical cyclone activity. Bull. Amer. Meteor. Soc., 85, 353–365Hanley, D., J.

Hall, J. D., A. J. Matthews, and D. J. Karoly, 2001: The Modulation of Tropical Cyclone Activity in the Australian Region by the Madden-Julian Oscillation. Mon. Wea. Rev., 129, 2970-2982. Hendricks, E. A., M. T. Montgomery, and C. A. Davis, 2004: On the role of “vortical” hot towers in formation of tropical cyclone Diana (1984). J. Atmos. Sci., 61, 1209–1232.

Heymsfield, G. M., J. B. Halverson, J. Simpson, L. Tian, and T. P. Bui, 2001: ER-2 Doppler Radar Investigations of the Eyewall of Hurricane Bonnie during the Convection and Moisture Experiment-3. J. App. Meteor., 40, 1310-1330. Hirschberg Paul A. and J. Michael Fritsch 1993.On Understanding Height Tendency Mon.Wea. Rev. Vol. 121, No. 9, pp. 2646–2661.

Holland,G.J.1984. On the climatology and structure of tropical cyclones in the Australian/southwest Pacific region. Aust. Met. Mag., 32, 1-46.

Holliday, Charles R. and Aylmer H. Thompson 1979: Climatological Characteristics of Rapidly Intensifying Typhoons, Monthly Weather Review: Vol. 107, No. 8, pp. 1022–1034.

Li Tim and Bing Fu. 2006: Tropical Cyclogenesis Associated with Rossby Wave Energy Dispersion of a Preexisting Typhoon. Part I: Satellite Data Analyses. Journal of the Atmospheric Sciences: Vol. 63, No. 5, pp. 1377–1389.

Li Tim and Bing Fu. 2006: Tropical Cyclogenesis Associated with Rossby Wave Energy Dispersion of a Preexisting Typhoon. Part II: Numerical Simulations. Journal of the Atmospheric Sciences: Vol. 63, No. 5, pp. 1390–1409.

Molinari, J., and D. Vollaro, 2000: Planetary and synoptic scale influences on eastern Pacific tropical cyclogenesis. Mon. Wea. Rev., 128, 3296-3307.

304

Molinari, J., D. Vollaro, and K. L. Corbosiero, 2004: Tropical storm formation in a sheared environment. J. Atmos. Sci., 61, 2493–2509.

Molinari, J,Peter Dodge, David Vollaro and Kristen L. Corbosiero Marks Frank Jr. 2006: Mesoscale Aspects of the Downshear Reformation of a Tropical Cyclone. Journal of the Atmospheric Sciences: Vol. 63, No. 1, pp. 341–354.

Montgomery, M.T., M. E. Nicholls, T. A. Cram, and A. B. Saunders. 2006:A Vortical Hot Tower Route to Tropical Cyclogenesis Journal of the Atmospheric Sciences: Vol. 63, No. 1, pp. 355–386. Nakazawa, T., 2001: Suppressed Tropical Cyclone Formation over the Western North Pacific in 1998. J. Meteor. Soc. Japan, 79, 173-183. Paterson, Linda A. Barry N. Hanstrum, Noel E. Davidson, and Harry C. Weber. 2005: Influence of Environmental Vertical Wind Shear on the Intensity of Hurricane-Strength Tropical Cyclones in the Australian Region Monthly Weather Review: Vol. 133, No. 12, pp. 3644–3660. 131B137.

Reasor, P. D., M. T. Montgomery, and L. Bosart, 2005: Mesoscale observations of the genesis of Hurricane Dolly (1996). J. Atmos. Sci., 62, 3151–3171

Ritchie, E. A., 2003: Some aspects of midlevel vortex interaction in tropical cyclogenesis. Cloud Systems, Hurricanes, and the Tropical Rainfall Measuring Mission (TRMM), Meteor. Monogr., No. 51, Amer. Meteor. Soc., 165–174.

Saïdou Moustapha Sall and Henri Sauvageot. 2005: Cyclogenesis off the African Coast: The Case of Cindy in August 1999 Monthly Weather Review: Vol. 133, No. 9, pp. 2803–2813.

Yuh-Lang Lin, Katie E. Robertson, and Christopher M. Hill. 2005: Origin and Propagation of a Disturbance Associated with an African Easterly Wave as a Precursor of Hurricane Alberto (2000) Weather Review: Vol. 133, No. 11, pp. 3276–3298.

Zipser, E. J., 2003: Some views on “hot towers” after 50 years of tropical field programs and two years of. TRMM data. Cloud Systems, Hurricanes, and the Tropical Rainfall Measuring Mission (TRMM), Meteor. Monogr., No. 51, Amer. Meteor. Soc., 49–58.

305

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES Topic 2.4 : Observing and Forecasting of Extratropical Transition Rapporteur: J. L. Evans The Pennsylvania State University 503 Walker Building University Park, PA, 16802 USA Email: jle7@psu.edu Fax: 1-814-865-3663 Working Group: S. Aberson, J. Beven, A. Burton, R. Edwards, C. Fogarty, B. Hagemeyer, R. Hart, N. Kitibatake, R. McTaggart-Cowan, D. Roth, J. Sienkiewicz, S. Spratt, C. Velden 2.4.1 Introduction At IWTC-V in Cairns, Australia, researchers and forecasters agreed that a research program focusing on important physical characteristics associated with the extratropical transition (or “ET”) of tropical cyclones (TC) was needed. This research program should promote analysis of existing observational and modeling datasets to: (i) improve analyses and prediction of the structure changes, significant weather, and ocean impacts associated with ET; (ii) address uncertainties associated with numerical predictions of ET in the region of the storm; (iii) understand its far-field impacts; and (iv) coordinate any ET research program with existing programs to obtain detailed observations of the evolutionary structure of extratropical transition. In the past four years, progress has been made in all of these areas. Contributions related to the observation and forecasting of ET are reviewed here. 2.4.2 Towards an Operational Definition of ET Operational forecasting of ET remains a significant challenge. ET is still a developing research area and lacks a fully developed common language (Jones et al. 2003). Some researchers will refer to any TC that enters mid-latitudes as having undergone ET. To aid application of conceptual models, some forecasters seem to prefer to emphasize a distinction between systems that undergo significant “transformation” to a largely cold-core system (with associated structural changes) and those that undergo “capture” by mid-latitude westerlies while retaining a largely warm-cored structure. Development of consistent descriptors and definitions of the stages of ET will help progress toward operational forecasting techniques. Any definition of ET should not only be precise enough to satisfy the needs of the operational and research communities and should be accessible to the general public as well. The lack of a concrete definition of ET that can be communicated to the general public can result in tragedy when tropical cyclone warnings are discontinued (due to the ET of the storm) but the storm makes landfall with potentially devastating societal consequences. To emphasize the importance of these storms, the Meteorological Service of Canada labels storms that have undergone ET as “post-tropical” and continues to use the storm name issued by NHC – thus, a storm such as Floyd (1999) moving into Canadian waters would be referred to as “Post-Tropical Cyclone Floyd.” Other governments have begun to consider this problem, but no uniform approach to communicating the dangers from extratropically transitioning storms to the public has been agreed among all affected nations.

306

Improved understanding of the processes through which ET occurs will aid in the development of a definition of ET. Thus, observational studies of ET are critical. Recent observational studies of transitioning tropical cyclones are reviewed in the next section. 2.4.3 Observations of ET Due to their devastating societal consequences, a few ET case studies were published prior to the 1950s (e.g., Pierce 1939; Palmén 1958). In the 1950s Matano and Sekioka began their systematic assessments of the surface signatures of a number of transitioning storms that impacted Japan (Sekioka 1956; Matano 1958). However, intentional field studies of the process of transition would have to wait for the twenty-first century! The results of two of these ET-focused studies of Hurricanes Michael (2000) and Ophelia (2005) are reviewed here. 2.4.3.1 Hurricane Michael (2000) Following the first International Workshop on Extratropical Transition (Jones et al. 2003), and knowing that the region around the Atlantic Provinces of Canada has the climatological peak for Atlantic ET events (Hart and Evans 2001), the Meteorological Service of Canada began planning for a field project on ET (Abraham et al. 2002; 2004). Their opportunity came with the ET of Hurricane Michael in October 2000. The Canadian National Research Council (NRC) Convair 580 (CV580) aircraft was used to conduct the first reconnaissance flight tasked for an ET event into Hurricane Michael as it underwent transition on 19 October 2000 while translating rapidly northeastward to the south of Newfoundland. On the synoptic scale, the strongest winds were to the southeast side of the storm within the precipitation region, with a strong (speed > 70 ms-1) southwesterly jet in the 500-2000 m layer. The precipitation shield extended roughly 250 km to the northwest and 135 km to the east of the storm center. This asymmetric synoptic structure is consistent with the right-of-track asymmetry in the winds and the left-of-track shift in precipitation typically observed during ET. Dropsondes, radar, and in situ microphysics measurements of temperature, wind, and cloud structures were obtained for the first time. A radar cross-section just south of the storm center revealed cloud tops near 11.5 km to the west of the center, and between 6.0 and 9.5 km east of the center. The southwest-to-northeast “tilt” of the storm center suggested by these cloud top differences was confirmed in a number of ways: first, the cloud-top circulation center was significantly displaced relative to the surface low pressure center to the southwest. The implied tilt was in excess of 80° from vertical. In addition, dropsonde wind speed and potential temperature cross-sections had a similar tilt in the wind and thermal fields. Dropsonde data also revealed relatively dry, low-level air to the northwest of the storm (even in the precipitation region). Within the jet on the southeast side of the storm, equivalent potential temperatures ( eϑ ) were similar to those in the extratropical air mass, which indicated that this air was being entrained into Michael. However, the low-level center still had relatively high values of eϑ consistent with tropical air. This warm air mass was most narrow near the surface and fanned outward with increasing altitude. Microphysical measurements of the precipitation region to the northwest of the center revealed deep, horizontally uniform, stratiform clouds with no embedded convection detected. In contrast, the cloud properties east of the surface low were indicative of embedded convection, but this convection was neither strong nor deep. Exceedingly high ice water contents (and high concentrations of small ice particles) in the clouds to the west of the storm center may have important implications for the precipitation formation mechanisms and efficiency during ET. Problems and limitations identified during the Michael flight enabled improvements to the data collection and flight planning for future missions. Since Michael (2000), the CV580 has provided the Canadian Hurricane Center with valuable data from near-land systems for Tropical Storm Karen (2001),

307

the remnants of Hurricane Isabel (2003), and for Hurricane Juan (2003) as it made landfall near Halifax, Nova Scotia. 2.4.3.2 Hurricane Ophelia (2005) On 16 and 17 September 2005, the NOAA U.S. Hurricane Research Division and the Canadian Meteorological Center conducted joint observational missions into Hurricane Ophelia as it was undergoing ET between Cape Hatteras, Virginia and Nova Scotia, Canada. One NOAA P-3 aircraft, a U. S. Air Force C-130-J aircraft, and an Aerosonde participated in the mission on 16 September, whereas only the P-3 gathered observations on 17 September. The P-3 released dropwindsondes in the storm environment, as well as inside and around the core of Ophelia, on both days. High resolution flight-level data, SFMR surface wind data, Doppler radar data and digital camera footage were also gathered on these flights. Special land-based rawinsondes provided by the U. S. National Weather Service were also incorporated into the post-analyses of Ophelia. The goal of the research missions was to document the structure of a tropical cyclone as it underwent ET, and to assess the impact of observations on subsequent numerical forecasts, including verifying the impacts far downstream. These were the first joint U.S.-Canadian missions to observe a system undergoing ET, and also the first successful penetration of, and return from, a tropical cyclone by an unmanned vehicle.

Fig. 2.4.1 700-hPa winds from 16 September 2005 research flights during the ET of Hurricane Ophelia (courtesy Sim Aberson, NOAA/HRD). Dropwindsonde and rawinsonde observations (Fig. 2.4.1)in the environment around Ophelia on 16 September revealed a strong mid-level southwesterly flow that was beginning to impinge on the tropical circulation and a low- to mid-level pressure center over Southern Ontario (about 1200 km to the northwest of Ophelia). By the next day, this mid-level low was over Maine, and a mid-level jet was flowing over the low-level circulation of Ophelia from southwest to northeast, which resulted in only a

308

shallow circulation below. This evolving tilt of the core is reminiscent of that of Michael on 19 October 2000. Dropwindsonde observations in the core of Ophelia on 16 September showed that the center of Ophelia was nearly vertically aligned, with warm, moist air in the center. Dry air was beginning to impinge on the core in the mid-levels (above 700 hPa). By the next day, the core was still moist. However, the thermal soundings confirmed that the center was strongly tilted from southwest to northeast as a result of the mid-level jet and associated sheared flow impinging on the Ophelia circulation. Airborne doppler radar winds at 1800 UTC on 16 September provide further evidence of the northeastward tilt of the center from the surface through the mid-levels, then back to the southwest aloft, with maximum horizontal displacements of the center of just a few km. By 17 September, Ophelia had so few precipitation scatterers that Doppler analyses were not possible. The inner-core data suggest a complicated structure on 16 September. Aircraft at three different levels in the vortex reported very different structures. The NOAA P-3 at about 14,000 feet reported moderate turbulence and a radius of maximum winds of just a few km. The Air Force C-130 reported a smooth flight at 10,000 feet with a much larger radius of maximum winds. Flying at 5,000 feet, the aerosonde did not penetrate to the center, but flew within the radius of maximum winds. Radar reflectivity at 1800 UTC 16 September from the NOAA P-3 exhibited a small comma-shaped feature resembling an eyewall open to the southern side, and a second band of convection to the northwest of the center. Just three hours later, the eyewall structure had deteriorated and the band to the northwest had weakened. The convection tilted upwind with height relative to the cyclonic flow in the center, but was nearly vertical in the convective band to the northwest. The convection near the center weakened with height above 1 km, whereas the convection in the band to the northwest had highest reflectivity at about 3 km altitude. By 2100 UTC, the tilt of the center was in the same direction, but the vertical displacements had nearly tripled. At mid-levels, multiple wind centers were now analyzed. The convection had disappeared by 17 September, which left a region of stratiform precipitation swirling in the low-level flow. 2.4.4 Current Methods for Forecasting the Timing of ET Shortly after IWTC-V, Hart (2003) developed the cyclone phase space diagram and Evans and Hart (2003) clearly demonstrated its utility as an intuitive and objective way to define the onset and completion of the ET process. This tool highlights the structural changes of the vortex and the near-storm environment and has proved to be an excellent and relatively objective method for determining the current state of the cyclone, as well as for intercomparing the forecast evolution of the cyclone in available numerical model guidance. The accessibility of this tool has led to its use in operational centres in the U.S., Canada, and Australia. While the cyclone phase space diagram has gained wide acceptance, its current dependence on model analyses argues against its use in isolation. Most operational centers indicate that they combine the cyclone phase space diagram with satellite guidance and conceptual models of ET (e.g., Foley and Hanstrum 1994; Fogarty 2002; Abraham and Bowyer 2002; Hart et al. 2006b). Blending of these fairly independent forms of guidance to create a forecast remains difficult and subjective. 2.4.5 Current Methods for Forecasting ET Impacts A variety of specific forecasting challenges are associated with ET. Of primary concern is the reintensification – or, in some cases, (e.g., Alberto 2006) intensification – of the storm as an extratropical system. This remains a difficult problem that must be dealt with by forecasters. Another major operational concern is the changes to the surface circulation both during the ET process and once transition is complete. Redistribution of the surface winds alters the pattern of damaging winds

309

over land and also drives ocean surface wave evolution. In addition to modifying the surface wind field, the structural changes in the storm during ET also lead to coincident modifications in the precipitation distribution associated with the transitioning cyclone. Finally, the possibility of tornadoes associated with an ET event must be considered during the tropical phase of these systems. 2.4.5.1 Surface Wind Structure As yet, no objective product exists to provide specific guidance on the wind field redistribution and (re-) intensification during ET. However, useful operational guidelines for changes in wind structure during ET have been determined from past events (e.g., Fogarty 2002). During ET, the maximum winds generally become displaced to the equatorward side (right NH; left SH) of the storm track and the surface pressure field spreads near the storm center, which results in an expansion of the outer wind radii and an increase in the storm size. Due to the rapid storm translation speed, the right-left asymmetry across track in the peak winds is emphasized during ET. In the later stages of ET, a horseshoe shaped wind maximum in the front half of the storm domain is observed in some cases (e.g. Edson 2004). Since the storm may decouple from the surface over the colder marine boundary layer (e.g., Abraham et al. 2004), the observed surface winds may be lower than expected. However, the degree to which decoupling occurs above the boundary layer is not readily apparent in individual events. Operational NWP may be unreliable in forecasting broad measures of cyclone structure through ET (e.g. Evans et al. 2006) as well as the downstream cyclone development (Anwender et al. 2006), so alternate forecast strategies must be sought (Jones et al. 2003). Improved use of ocean surface winds, particularly from the NASA QuikSCAT satellite, has allowed better monitoring of the changes in the cyclone wind field as ET occurs. In a study of factors affecting gale radius evolution in warm-season cyclones, Higgs (2005) demonstrated that the storm and environmental modulators of cyclone wind structure differ for tropical and extratropical cyclones. Although two datasets (Kimball and Mulekar 2004; Moyer and Evans 2006) of significant wind speed radii in tropical cyclones currently available are internally consistent (statistically and theoretically), some systematic differences exist between these datasets (Moyer and Evans 2006). Thus, validation of products designed to forecast such surface wind changes is problematic. 2.4.5.2 Wave Field Evolution Transitioning tropical cyclones can become very efficient ocean surface wave producers. Strong winds to the right of the track blowing in the same direction as the storm motion can create very high seas where “trapped-fetch” resonant wave growth occurs (Bowyer and MacAfee 2005). As the storm accelerates and the wind field expands during ET, the wave maximum becomes displaced farther and farther to the right of the storm track. The arrival of the wave maximum at a location typically lags the passage of the storm by 2-3 hours. An operational modeling tool has been developed recently to compute dominant wave trajectories and significant wave heights for TCs undergoing the rapid forward translation typical of an ET event (MacAfee and Bowyer 2005, 2006). Wave forecasts from this model are available within less than one minute after the forecaster produces or changes the forecast track. This technique is not yet implemented in all regions affected by ET events. 2.4.5.3 Precipitation Distribution Changes Heavy rainfall associated with an ET event is usually concentrated along quasi-stationary frontal zones well ahead of the transitioning storm and within a few hundred kilometers to the left of the storm track in the “delta” rain region identified by Shimazu (1998). This heavy precipitation zone is associated with the tropical cyclone outflow extending poleward from the cyclone center (Kitabatake 2002). Due to the

310

intrusion of dry air wrapping around the southern part of the storm, precipitation is usually substantially less to the right of the storm track. Due to the import of moist tropical air to higher latitudes, the potential exists for extreme precipitation in flash floods associated with upslope flow in the vicinity of mountainous regions. Beginning from these general guidelines, forecasters at the U.S. Hydrometeorological Prediction Center (HPC) have developed a subjective methodology for forecasting the evolution of the precipitation distribution as a storm undergoes ET (Roth 2006). When it appears that a tropical cyclone is becoming extratropical and/or fronts have moved relatively close to the storm center, the forecast region of heavy precipitation is shifted from right to left of track. While this “rule of thumb” is typically effective, storms such as Wilma (2005) and Alberto (2006) demonstrate that exceptions to this rule remain to be explained. To provide guidance on more refined spatial detail in the quantitative precipitation forecast (QPF), the operational model track forecast that is closest to the official Tropical Prediction Center (TPC) track forecast is identified. The regions of strongest flow are identified that are: (i) in excess of 35 kt; and (ii) perpendicular to a frontal or coastal boundary or local terrain. Heavy precipitation is expected in these regions due to strong forced ascent. As with standard QPF products, favorable quadrants of upper-level jet streaks, areas of frontogenesis, and potential vorticity anomalies are also used as indicators of local QPF maxima. Conceptual models and current precipitation structure (determined from radar and satellite) are used to ensure reasonable spatial continuity in the heavy precipitation region. The diurnal cycle of precipitation will enhance core rainfall overnight and outer rainbands during day. Storm analogs are also a useful check of a QPF forecast. Selection of an analog depends on satisfying a number of common criteria: (i) size of the current rain shield; (ii) vertical wind shear; (iii) similar storm track with similar proximity to topography; and (iv) fronts in the vicinity of the storm. An important limitation is that not all events will have a useful analog. Past events may also highlight locations susceptible to strong topographic rainfall enhancement (Roth 2006). Past events are also used to calibrate the likely average or extreme QPFs based on observed tropical cyclone impacts over the past 15-25 years. Finally, checks are applied on the upper limits of QPF amounts, both for areal average amounts in the QPF graphics and the text QPF statement to TPC. For example, a soft cap of 2.5 inches per six-hour period is usually enforced for the 28-km forecast grid. 2.4.5.4 Tornadoes Associated with ET Studies of Florida tornadoes associated with tropical and hybrid cyclones (cyclones having both tropical and extratropical characteristics) reveal an increased likelihood of tornadoes in hurricanes interacting with the mid-latitudes (Hagemeyer 1997, 1998). Since the hybrid cyclones studied were documented to be interacting with a midlatitude trough to the northwest and to be accelerating to the north or northeast, these storms are at least potentially beginning ET and may provide a useful model for tornado forecasting in other ET events. While tornadoes associated with the passage of a hybrid cyclone are consistently the most dangerous (all resulted in injuries and most were “killer” tornadoes), they result in longer and more active tornado outbreaks than from purely tropical cyclones and are also the rarest form of warm-season cyclone-related tornadoes documented in Florida (Hagemeyer 1997). Most of these documented tornadoes occurred in the overrunning zone very near or north of where a strong low-level jet (> 35 kt or 17.5 ms-1) intersected the surface warm front, maximizing moisture convergence in a highly sheared, moist environment with diffluence at high levels. In each case, the heavy rain phase began before the severe weather phase and continued during, and well after, the tornado outbreak phase. The tornadic phases appeared to be associated with the approach of an upper short wave, an increase of the 850-500 mb wind speeds, and maximum shear and overrunning (Hagemeyer 1997). Hagemeyer (1998) presents general guidelines (from climatological analyses of tornadic outbreaks in

311

Florida) for forecasting tornadoes related to the passage of a tropical or hybrid cyclone. These tornadoes tend to be stronger than "typical" Florida tornadoes. June, September and October are peak months for tornadoes associated with either tropical or hybrid cyclones, with significant hybrid cyclone tornadoes most likely in June and October (note that October is climatologically the month with the highest percentage of ET events; Hart and Evans 2001). Fast-moving tropical and hybrid cyclones are more likely to produce significant tornadoes. Outer rainbands of tropical cyclones with discernable hybrid characteristics are generally more likely to produce significant tornadoes. Curtis (2004) discovered that 11 of 13 tornado outbreak cases associated with landfalling TCs along the Atlantic coast and Gulf of Mexico since 1960 contained clear evidence of a dry intrusion at midlevels over the outbreak area. Two distinct patterns were identified with respect to the source of the midlevel dry air. In one, a mass of dry air that impinged on much of the northern or northwestern semicircle of the storm’s outer circulation became divided into two lobes as the storm advanced, with one lobe to the northwest and the other to the northeast of the storm center. The other pattern involved ingestion of a lobe of dry air from a reservoir most often (but not exclusively) located in the eastern semicircle of the storm. Each pattern suggests the role of baroclinic processes (or increasing hybrid nature) in aiding tornadogenesis. Significant tornadoes are most likely in association with higher reflectivity cells contained within dominant outer rainbands in the right-front quadrant of north- to northeast-moving tropical or hybrid cyclones in the Gulf of Mexico, regardless of central convection and central pressure. That is, a monotonic relationship does not exist between tornado strength and cyclone intensity (Hagemeyer 1998). 2.4.6 Recent Research Relevant to ET Forecasting The ET forecast problem seems to be nearly separable into three issues: (i) the evolution of significant weather and waves associated with the storm; (ii) the possible rapid reintensification of the primary vortex; and (iii) the downstream propagation of Rossby waves generated by perturbing the midlatitude flow by the transitioning storm. As discussed above, studies of the first issue are advancing, mainly in operational centers (e.g. Roth 2006; Bowyer and MacAfee 2005). 2.4.6.1 Evolution and Simulation of the ET Vortex The relative sensitivity of the ET process to the TC structure versus the midlatitude environment into which the storm moves has been an important focus of operationally-relevant research since IWTC-V. Recent studies on this and associated topics have engaged the problem from different perspectives (Agustí-Panareda et al. 2004; Evans and Prater-Mayes 2004; McTaggart-Cowan et al. 2004; Morgan 2004, 2006). Although each study in isolation is unable to make a general assessment of the importance of the remnant TC vortex versus the midlatitude flow, taken as a whole, this growing body of literature promises to aid in the development of conceptual models and forecasting tools that will be of significant utility to forecasters in the near-future. The numerical modeling studies of Evans and Prater-Mayes (2004) and Ma et al. (2006) highlight the sensitivity of the numerical representation of ET to details of the model configuration and initialization. Such model sensitivities were mooted by Jones et al. (2003). Further documentation for the impact of model initial conditions on the simulation of the structure of an evolving ET is provided in Evans et al. (2006), who compare the structural evolution of the evolving storm as represented in the cyclone phase space of Hart (2003). Rather than requiring exact replication of analyzed and forecast structure, they follow Arnott et al. (2004) and cluster the identified structures into seven groups: three tropical cyclone types of increasing intensity; two hybrid forms (including the transitioning phase); and two extratropical. They find that the NOGAPS model simulations had the best 12-24 hour structure forecasts, but that the imposition of the synthetic vortex in the NOGAPS initial conditions resulted in less useful forecasts at

312

36 hours. In contrast, the most recent formulation of the GFS model (with vortex relocation only) was most useful at the 36-hour lead time.

Fig. 2.4.2 Sea-level pressure (every 4 hPa) for Hurricane Michael (2000) valid at 0000 UTC 20 October 2000 based on: (a) subjective, manual analysis; (b) 24-hour GEM regional forecast; (c) 24-hour “no-vortex” simulation of the 12-km MC2 model; and (d) 24-hour simulation of the control run of the MC2 model with vortex insertion employed (courtesy of Chris Fogarty, MSC/CHC). These results are consistent with the recent work by Fogarty et al. (2006a, b), who have demonstrated the utility of using synthetic vortex insertion in the initial pre-ET atmospheric fields during Hurricanes Juan (2003) and Michael (2000). One of the major problems with forecasting Hurricane Michael was that some numerical models were favoring the development of the baroclinic low off the coast of Nova Scotia rather than the transitioning storm that actually remained the primary low pressure center (Abraham et al. 2004). The Fogarty et al. (2006b) simulations of Hurricane Michael (2000) demonstrate that accurate specification of the tropical vortex prior to ET leads to a more faithful reproduction of the storm evolution with the baroclinic cyclone (Fig. 2.4.2). Similar to Evans et al. (2006), Fogarty et al. (2006a, b) also conclude that the use of a synthetic vortex in the initialization procedure should be limited to forecast periods less than 2-3 days and prior to ET.

313

Whereas the composite analyses of Arnott et al. (2004) emphasized the key, common features of a North Atlantic cyclone undergoing ET, those of Hart et al. (2006b) delineate the different synoptic factors determining post-ET evolution. The relatively small spread of trajectories in the tropical sector of the cyclone phase space contrasted markedly with the increase in trajectory variability once ET had begun, which provided justification for stratification of these 34 cases by post-ET evolution. The results of these analyses provide a basis for developing testable forecast rules for post-ET storm evolution. For example, post-ET intensifiers begin transition with a negatively tilted trough ~1000km upstream, while post-ET weakeners commence transition ~1500km east of a positively tilted trough. Hart et al. (2006b) hypothesize that the negative trough tilt in the intensifying cases permits a contraction and intensification of the eddy potential vorticity flux, while the positive tilt associated with the weakeners prevents contraction and intensification of the forcing. Identification of the relatively few ET events (6/34 in their study) that will undergo warm-seclusion is critical since such systems greatly increase the chance of post-ET wind and wave damage. Hart et al. (2006b) found that such an evolution is most likely when the scales of the interacting trough and the transitioned ET vortex are similar. 2.4.6.2 Downstream Impacts of the ET Vortex Recent investigations have begun to address the influence of the tropical cyclone anticyclonic outflow and tropical moisture content on the excitation of standard and diabatic Rossby waves. For example, McTaggart-Cowan et al. (2006) demonstrate that a strong tropical cyclone that does not undergo significant reintensification during ET can still introduce a substantial perturbation – with associated forecast uncertainty – to the extratropics, and even back into the tropics. The growth of model error associated with the ET event has been shown by Harr (2006) and Riemer (2006) to propagate downstream at a speed between the phase speed and group velocity of the wavetrain generated. Harr (2006) further used EOF analysis and clustering to begin the development of a possible forecasting tool capable of identifying likely scenarios associated with individual ET events. Anwender et al. (2006) considered the impact of perturbations to the tropical vortex on ensemble forecasts of the downstream weather impacts of an ET event. 2.4.6.3 Potential Role of ET in Seasonal Variability Beyond the regional and medium-range hemispheric impacts of ET, Hart (2006) has begun to investigate the relationship between recurving tropical cyclones and the subsequent winter climate. Although this work is still in its relatively early stages, it is interesting to consider the impact of anomalous recurving tropical cyclone frequency on the variability in the hemispheric meridional temperature flux at 500 hPa (Fig. 2.4.3). Preliminary results suggest that anomalous TC recurvature leads to anomalous snowcover at the start of winter that does not return to normal by winter's peak, thereby altering the albedo of the hemisphere and the radiative balance (Hart et al. 2006a).

314

Fig. 2.4.3 The JFM zonal-mean 500mb temperature flux following five or less Northern Hemisphere recurving tropical cyclones (red) and twelve or more (blue). Top panel: 1955-2005. Bottom panel: 1979-2005. Quartiles 1 and 3 are the same for the two periods. Source of data: NCEP/NCAR reanalysis (Kalnay et al. 1996). Figure reproduced from Hart et al. (2006a). 2.4.6.4 Development of Forecast Diagnostics of Potential Interest to the Forecast Community Since 2004, researchers at the State University of New York at Albany have been generating realtime diagnostic displays, primarily for map room discussions(http://www.atmos.albany.edu/facstaff/rmctc/DTmaps/animSelect.php.The animation-ready images are archived for approximately three months. Images are generated from the high-resolution (0.5 degree) GFS final global analysis using diagnostics computed on the sphere for enhanced accuracy. The system has been designed to be flexible to make it simple to add new images, domains, and levels upon request. During periods of enhanced tropical/extratropical interaction (such as ET events), streamfunction, velocity potential, and quasi-geostrophic (QG) diagnostics are of particular interest. The first two quantities are useful because the flow at all latitudes is faithfully represented by the wind field components. The streamfunction field reveals the meridional extent of the midlatitude features, while the velocity potential highlights regions of large-scale ascent either driven by dynamical forcing or persistent convection. Two sets of QG diagnostics are provided: a traditional set computed using the geostrophic winds; and a second set based on the non-divergent wind field. This second method for computing the Q-vector fields has the advantage of minimizing noise in the tropics and near active convection. The intention is to use these fields to estimate the strength of the forcing that the large-scale flow is applying in the vicinity of the tropical system during both ET and tropical transition, which is the transformation of a cold-core baroclinic system to a tropical cyclone. The working hypothesis is that when the non-divergent Q-vector ascent forcing reaches a maximum near the storm center, significant constructive interaction will occur between the trough and the tropical feature. These

315

diagnostics are still under development and review at the State University of New York at Albany, although they are freely available at the URL noted above. 2.4.6.5 ET climatology The climatology of Hart and Evans (2001) documented the importance of ET in the North Atlantic. However, this study lacked regional detail. A climatology of tropical cyclones that have affected Canada since 1900 is presently under construction. This climatology will include the mean wind and rainfall distributions throughout various stages of ET, and the case studies of meteorological fields and impacts from historical events are intended to serve as a reference for forecasters (Fogarty, pers. comm. 2006). Detailed climatologies of this type that also incorporate the range of impacts of historic ET events would be of value in all regions affected by ET. 2.4.7 Roadblocks for further advancements in ET forecasting

Several pressing needs should be addressed in the future for ET diagnosis and forecasting:

(i) Improve forecasts of the timing of ET.

Given the uncertainty in current methods for identifying exactly when ET is taking place, and the importance of structure changes associated with the transition on downstream forecasts, it is imperative that research efforts focus on better observations of the event with a goal to refining conceptual models of the transition process.

(ii) Improve the analysis of storm intensity during ET. At the Third International Workshop on Extratropical Transition in Perth during 2005, a need was identified to direct research toward the development of a satellite-based method to better observe/diagnose the ET process. One potential approach is development of a satellite-based ET identifier. Since the Dvorak IR-based method does not perform adequately as an intensity estimation tool on systems undergoing ET, some operational forecast centers have attempted local modifications that typically use only single channel analyses. Observed ET signatures from SSMI, TRMM, and AMSU instruments, or products such as high-resolution cloud-tracked winds, are all promising. However, no unified multi-spectral approach to observing ET has yet been developed. Such a tool would provide useful operational guidance.

(iii) Improve the forecast of cyclone intensity during ET, as well as verification of these forecasts.

(iv) Improve the forecast of tropical cyclone size, particularly the radii of the 34-kt, 50-kt, and

64-kt winds that are used in TC advisories and non-tropical warnings, as well as provide verifications of these forecasts; These issues may be helped by improved numerical weather prediction. However, ET-specific statistical-dynamical techniques analogous to the tropical cyclone forecast techniques such as SHIPS and the wind radii CLIPER may be a first step.

(v) Improve the cyclone phase space diagram by direct use of observations instead of model analyses.

As a proof-of-concept, satellite-based analyses of specific case studies using microwave soundings have been performed. However, the satellite-based cyclone phase space is far from being operational.

316

(vi) Improve guidance for tornado outbreaks associated with a landfalling TC (whether ET or not).

(vii) Improved guidance for the precipitation structure during ET.

Explorations of the cyclone phase space as a diagnostic for surface wind radii and rainfall transitions are presently underway. Interestingly, forecasting roadblocks identified here have much in common with the forecast needs given in Jones et al. (2003). 2.4.8 Proposals for moving forward An improved understanding of the structural changes and impacts during the extratropical transition from a tropical cyclone to an extratropical cyclone will contribute to the development of improved conceptual and numerical models that will enable weather forecasters to better anticipate changes, and improve warnings associated with ET (Abraham et al. 2004). Such understanding could be gained through post-analyses of challenging forecast cases (e.g., Maria and Wilma from 2005) and focused field experiments. Field experiments in planning stages for THORPEX (such as the Pacific Asian Regional Campaign, or PARC, presently slated for 2008; Shapiro and Thorpe 2005) provide ideal fora for exploring the storm-scale evolution (including intensity, structure, and wave impacts), downstream predictability associated with ET, and for testing forecasting paradigms. Determining the nature of the indirect effects of ET, and the physical mechanisms behind them, is an important current research topic of relevance to the forecast community. Higher resolution global analyses that assimilate an increasing number of observations will help to improve our understanding and aid in the creation of conceptual models of the impacts of ET on the larger-scale flow. Evaluations of the extent and multi-scale nature of the interactions between the remnant TC and its environment from diagnostic, modeling, and ensemble-based perspectives are promising avenues for forecast improvement. Improvements in the availability of forecast tools and analyses to forecasters across all affected regions, and continued communication of recent results between the operational and research communities, remain vital to the advancement of ET forecasting. Bibliography Abraham, J., and P. Bowyer, 2004: Hurricanes, Canadian style: Extratropical transition. UCAR COMET module. Available at http://www.meted.ucar.edu/norlat/ett/index.htm. Abraham, J., C. T. Fogarty, and W. Strapp, 2002: Extratropical transition of Hurricanes Michael and Karen: Storm reconnaissance with the Canadian Convair 580 aircraft. 25th AMS Conference on Hurricanes and Tropical Meteorology, 29 April-3 May 2002, San Diego CA. Abraham, J., W. Strapp, C. Fogarty, and M. Wolde, 2004: Extratropical transition of Hurricane Michael: An aircraft investigation. Bull. Amer. Meteor. Soc., 85, 1323-1339. Agustí-Panareda, A., C. D. Thorncroft, G. C. Craig and S. L. Gray, 2004: The extratropical transition of hurricane Irene (1999): A potential vorticity perspective. Quart. J. Roy Meteor. Soc., 130, 1047-1074. Anwender, D., M. Leutbecher, S. Jones, and P. Harr, 2006: Sensitivity of ensemble forecasts of extratropical transition to initial perturbations targeted on the tropical cyclone. 27thConference on

317

Hurricanes and Tropical Meteorology, 24-28 April 2006, Monterey, CA., Amer. Meteor.Soc. Arnott, J. M., J. L. Evans, and F. Chiaromonte, 2004: Characterization of extratropical transition using cluster analysis. Monthly Weather Review, 132, 2916-2937. Bowyer, P. J., and A. W. MacAfee, 2005: The theory of trapped-fetch waves with tropical cyclones – An operational perspective. Weather and Forecasting, 20, 229-244. Curtis, L., 2004: Midlevel dry intrusions as a factor in tornado outbreaks associated with landfalling tropical cyclones from the Atlantic and Gulf of Mexico. Weather and Forecasting, 19, No. 2, 411-427. Edson, R. T., 2004: Tropical cyclone analysis techniques from QuikSCAT NRCS, wind and ambiguity data and microwave imagery. 26th AMS Conference on Hurricanes and Tropical Meteorology, Miami, FL. Evans, J. L., J. M. Arnott, and F. Chiaromonte, 2006: Evaluation of operational model cyclone structure forecasts during extratropical transition. Monthly Weather Review, (in press). Evans, J. L., and R. E. Hart, 2003: Objective indicators of the life cycle evolution of extratropical transition for Atlantic tropical cyclones. Monthly Weather Review, 131, 909-925. Evans, J. L., and B. E. Prater-Mayes, 2004: Factors affecting the posttransition intensification of Hurricane Irene (1999). Monthly Weather Review, 132, 1355-1368. Fogarty, C. T., 2002: Operational forecasting of extratropical transition. 25thConference on Hurricanes and Tropical Meteorology, 29 April-3 May 2002, San Diego, CA, Amer. Meteor. Soc. Fogarty, C. T., R. J. Greatbatch, and H. Ritchie, 2006a: The role of anomalously warm sea surface temperatures on the intensity of Hurricane Juan (2003) during its approach to Nova Scotia. Monthly Weather Review, 134, 1484-1504. Fogarty, C. T., R. J. Greatbatch, and H. Ritchie, 2006b: A numerical modeling study of the extratropical transition of Hurricane Michael (2000). Submitted to Weather and Forecasting. Foley, G. R., and B. N. Hanstrum, 1994: The capture of tropical cyclones by cold fronts off the west coast of Australia. Weather and Forecasting, 9, 577-592. Hagemeyer, B. C., 1997: Peninsular Florida tornado outbreaks. Weather and Forecasting, 12, 399-427. Hagemeyer, B. C., 1998: Significant tornado events associated with tropical and hybrid cyclones in Florida. 16th AMS Conference on Weather Analysis and Forecasting, 11-16 January 1998, Phoenix, AZ. Harr, P., D. Anwender, and S. C. Jones, 2006: Predictability associated with the downstream impacts of the extratropical transition (ET) of tropical cyclones. 27th Conference on Hurricanes and Tropical Meteorology, 24-28 April 2006, Monterey, CA., Amer. Meteor. Soc. Hart, R. E. and J. L. Evans, 2001: A climatology of extratropical transition of Atlantic tropical cyclones. J. Climate, 14, 546-564. Hart, R. E., 2003: A cyclone phase space derived from thermal wind and thermal asymmetry. Monthly Weather Review, 131, 585-616. Hart, R. E., 2006: The winter impact of recurving tropical cyclones. 27th Conference on Hurricanes and

318

Tropical Meteorology, 24-28 April 2006, Monterey, CA. Hart, R. E., L. F. Bosart, and C. Hosler, 2006a: The possible hemispheric impacts of anoamlous recurving tropical cyclone frequency. Submitted to Mon. Wea. Rev., August 2006. Hart, R. E., J. L. Evans, and C. Evans, 2006b: Synoptic composites of the extratropical transition lifecycle of North Atlantic tropical cyclones: Factors determining post-transition evolution. Monthly Weather Review, 134, 553-578. Higgs, J., 2005: Slice inverse regression and principal component analysis of factors affecting cyclone gale radius. Masters Thesis, Department of Meteorology, The Pennsylvania State University. Jones, S. C., P. A. Harr, J. Abraham, L. F. Bosart, P. J. Bowyer, J. L. Evans, D. E. Hanley, B. N. Hanstrum, R. E. Hart, F. Lalaurette, M. R. Sinclair, R. K. Smith, and C. Thorncroft, 2003: The extratropical transition of tropical cyclones: Forecast challenges, current understanding, and future directions. Weather and Forecasting, 18, 1052-1092. Kimball, S., and M. S. Mulekar, 2004: A 15-year climatology of North Atlantic tropical cyclones. Part I: Size parameters. J. Climate, 17, 3555-3575. Kitabatake, N., 2002: Extratropical transformation of Typhoon Vicki (9807): Structural changes and the role of upper-tropospheric disturbances. J. Meteor. Soc. Japan, 80, 229-247. Klein, P. M., P. A. Harr, and R. L. Elsberry, 2000: Extratropical transition of western North Pacific tropical cyclones: An overview and conceptual model of the transformation stage. Weather and Forecasting, 15, 373–396. Klein, P. M., P. A. Harr, and R. L. Elsberry, 2002: Extratropical transition of western North Pacific tropical cyclones: Midlatitude and tropical cyclone contributions to reintensification. Monthly Weather Review, 130, 2240-2259. Ma, S., H. Ritchie, J. R. Gyakum, J. Abraham, C. T. Fogarty, and R. McTaggart-Cowan, 2003: A study of the extratropical reintensification of former Hurricane Earl using Canadian Meteorological Centre regional analyses and ensemble forecasts. Monthly Weather Review, 131, 1342-1359. MacAfee, A. W., and P. J. Bowyer, 2005: The modeling of trapped-fetch waves with tropical cyclones — A desktop operational model. Weather and Forecasting, 20, 245–263. MacAfee, A. W., and P. J. Bowyer, 2006: Corregium. Weather and Forecasting, 21, 429. Matano, J., 1958: On the synoptic structure of Hurricane Hazel, 1954, over the eastern United States. J. Meteor. Soc. Japan, 36, 23-31. McTaggart-Cowan, R., L. F. Bosart, J. R. Gyakum, and E. H. Atallah, 2006: Evolution and global impacts of a diabatically-generated warm pool: Hurricane Katrina (2005). 27th AMS Conference on Hurricanes and Tropical Meteorology, April 2006, Monterey, CA. McTaggart-Cowan, R., J. R. Gyakum, and M. K. Yau, 2004: The impact of tropical remnants on extratropical cyclogenesis: Case study of Hurricanes Danielle and Earl (1998). Monthly Weather Review, 132, 1933-1951. Morgan, M., 2005: Adjoint-based sensitivity analysis of the (possible) extratropical transitions of Hurricanes Floyd (1999) and Earl (1998). 1st THORPEX International Science Symposium, December 2004, Montreal QC.

319

Morgan, M., 2006: Adjoint-derived forecast sensitivity of hurricane track and extratropical transition. 27th Conference on Hurricanes and Tropical Meteorology, 24-28 April 2006, Monterey, CA, Amer. Meteor. Soc. Moyer A., and J. L. Evans, 2006: A study of current datasets for outer wind radii. 27th Conference on Hurricanes and Tropical Meteorology, 24-28 April 2006, Monterey, CA, Amer. Meteor. Soc. Palmén, E., 1958: Vertical circulation and release of kinetic energy during the development of hurricane Hazel into an extratropical storm. Tellus, 10, 1-23. Pierce, C., 1939: The meteorological history of the New England hurricane of Sept. 21, 1938. Monthly Weather Review, 67, 237-288. Riemer, M., 2006: The impact of extratropical transition on the downstream flow: Idealized modeling study. 27th Conference on Hurricanes and Tropical Meteorology, 24-28 April 2006, Monterey, CA, Amer. Meteor. Soc. Roth, D. M., 2006: Tropical cyclone-related Quantitative Precipitation Forecasting at HPC. 60th Interdepartmental Hurricane Conference, 20-24 March 2006, Mobile, AL. Available at http://www.ofcm.gov/ihc06/linking_file_ihc06.htm Sekioka, M., 1956: A hypothesis on complex of tropical and extratropical cyclones for typhoon in middle latitudes, I. Synoptic structure of Typhoon Marie over the Japan Sea. J. Meteor. Soc. Japan, 34, 42-53. Shapiro, M. A., and A. J. Thorpe, 2005: THORPEX International Science Plan Version III. THORPEX International Programme Office, Atmospheric Research and Environment Programme Department, World Meteorological Organization Secretariat, Geneva, Switzerland, 51 pp. Available at www.wmo.int/thorpex. Shimazu, Y., 1998: Classification of precipitation systems in mature and early weakening stages of typhoons around Japan. J. Meteor. Soc. Japan, 76, 437-445.

320

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES Topic 2.5 : Physical Processes and Downstream Impacts of Extratropical Transition Rapporteur: John R. Gyakum Department of Atmospheric and Oceanic Sciences McGill University 805 Sherbrooke Street West Montreal, QC H3A 2K6 Canada E-mail: john.gyakum@mcgill.ca Fax: 514.398.6115 Working Group: L. F. Bosart, C. Fogarty, P. Harr, S. Jones, R. McTaggart-Cowan, W. Perrie, M. Peng, M. Riemer, R. Torn 2.5.1 Introduction Substantial progress in the understanding of extratropical transition (ET) has been made since the period of the last report written for the IWTC-IV, in which much of the material was derived from the review paper of Jones et al. (2002). The importance of the ET in influencing the dynamics of the atmosphere has become more evident in recent years. The occurrence of ET events over the North Pacific has been observed to coincide with periods of reduced forecast model skill (Jones et al. 2003). Harr et al. (2004) examined downstream propagation of increased ensemble standard deviation in mid-tropospheric heights from several operational global model ensemble prediction systems during the ET of typhoon Maemi (2003) that also coincided with reduced forecast skill over the Northern Hemisphere. The complex physical and dynamical processes during ET are extremely sensitive to sources and impacts of initial condition errors and forecast model uncertainty. Therefore, factors that impact forecast model error growth downstream of an ET event must be identified. For example, predictability during an ET event may exhibit large variations due to the phasing between the decaying tropical cyclone and the midlatitude circulation (Klein et al. 2002). McTaggart-Cowan et al. (2006a) have shown that the ET of Hurricane Katrina (2005), though lacking in reintensification, had an important impact on the middle-latitude flow and reduced medium-range predictability. The purpose of this discussion is to identify crucial physical processes associated with ET, and the associated downstream impacts. Specific forecasting issues and observation strategies to improve forecasting and understanding of the ET process are discussed elsewhere. 2.5.2 Physical Processes The poleward movement and ET of a tropical cyclone (TC) initiates complex interactions with the middle latitude environment that often results in a high-impact midlatitude weather system with strong winds, high seas, and large amounts of precipitation (Atallah et al. 2006). Although these extreme conditions severely impact the region of the ET, there are significant impacts downstream of the ET event, owing to the excitation of large-scale propagating Rossby wave-like disturbances. Although ET occurs over several ocean basins, the largest number of ET events usually occurs during the Northern Hemisphere summer and fall seasons over the western North Pacific (Jones et al. 2003). The case of supertyphoon Dale of 1996 (Kelsey and Bosart 2006) offers an excellent example of a western North

321

Pacific ET. Downstream impacts propagate across the North Pacific to impact North America via the forcing of anomalous cyclonic and anticyclonic circulations. Despite the large number of western North Pacific Basin ET events, important tropical characteristics of hurricanes are often maintained in higher latitudes of the North Atlantic, as in the case of Juan (2003) (McTaggart-Cowan et al. 2006b, c) and in even in the South Atlantic as in Caterina (2004) (McTaggart-Cowan et al. 2006d). Indeed, Dickinson et al. (2006) have documented a relatively rare case of ET in the eastern North Pacific, that of Lester (1992). Scientific issues associated with ET and downstream impacts due to ET events may be placed in a framework of mechanisms, predictability, and strategies for increasing predictability. The ET process may be characterized by complex physical interaction within three interrelated regions: the tropical cyclone core, the tropical cyclone-middle latitude interface, and the middle latitude impact region. To understand the impact of ET on high-impact downstream weather events, mechanisms responsible for the generation, intensification, and propagation of the Rossby wave-like disturbances need to be identified. All three regions of the ET process likely play important roles in the mechanisms responsible for downstream impacts due to ET. A Rossby wave response may be forced by advection of vorticity due to the divergent wind (Sardeshmukh and Hoskins 1988), which may result from the tropical cyclone core. A similar mechanism may be associated with diabatic Rossby waves (Moore and Montgomery 2005) due to upward motion along sloping isentropic surfaces that exist at the tropical cyclone middle latitude interface (Harr and Elsberry 2000). Finally, the middle latitude impact region provides the avenue by which the wave energy impacts the middle latitude circulation into which the decaying TC is moving. Furthermore, the downstream response to ET events exhibits large spatial and temporal fluctuations, which may be related to specific characteristics of each of the three ET regions. One ongoing project most relevant to physical processes during ET is the configuration, testing, and development of a limited area model in Canada (MC2) of the Global Environmental Multiscale (GEM) code with synthetic vortex insertion employed (pre-ET) to initiate short-range numerical forecasts designed to assist in forecast production. Validation of this model includes analysis of aircraft data collected primarily from dropsondes during a handful of storms since the year 2000. The aircraft component of this work was conducted by the National Research Council (NRC) of Canada using the Convair-580 aircraft (Wolde et al. 2001) organized by Jim Abraham (MSC) and Walter Strapp (MSC/NRC). An analysis of aircraft and dropsonde data outlining interesting physical characteristics of the rapidly-transitioning Hurrricane Michael in October 2000 can be found in Abraham et al. (2004). More recently, a collaborative observing study involving two flights into the ET of Tropical Storm Ophelia in September 2005 was organized by Jim Abraham of the MSC and Sim Aberson of the Hurricane Research Division (HRD) in Miami, Florida. Some interesting findings from those flights were presented in May 2006 at a poster session at the Canadian Meteorological and Oceanographic Society (CMOS) Congress in Toronto, Ontario, Canada (Fogarty, 2006). Primary findings from these aircraft flights indicated extreme boundary layer wind shears during the ET over cool sea surface temperatures off Eastern Canada, and a significant deepening of the usual low-level wind maximum in TCs. During the ET of Ophelia, the upper portion of the storm’s circulation was sheared away by strong upper level winds accompanied by the intrusion of dry, mid-latitude air. This may have influenced the representative steering layer for the storm, making it challenging to forecast its movement. Hurricane wind speeds were also observed only 200 m above the ocean surface to the right of the storm track. This was surprising given that the storm’s central pressure had risen to near 1000 hPa – but highlights the high wind potential to the right of seemingly benign ET events. The results of these aircraft studies are used at the Canadian Hurricane Centre (CHC) to understand the behavior of hurricanes undergoing ET. The aircraft/dropsonde data also aid in validation of the

322

experimental in-house model for hindcast simulations. The “hurricane configuration” of the MC2 model used for hindcast sensitivity studies (Fogarty et al. 2006a,b) shows promise as a forecast tool. Equally important is the acquired knowledge from these studies that provide 4-dimensional (3D plus time) imagery of the evolution of moisture and mass fields during the ET process. Program leads Peter Bowyer, Jim Abraham and Chris Fogarty share findings with staff through training sessions, workshops, and one-on-one discussions. Output from several of the experiments demonstrating structural changes taking place during ET can also be viewed at: http://projects.novaweather.net/work.html. An idealized modeling study used a dry three-dimensional primitive equation model to investigate the ET of tropical-cyclone-like vortices in developing baroclinic waves (Weindl 2004). The sensitivity of the ET to the initial location of the tropical cyclone was investigated for the two paradigms of baroclinic wave life cycles, one characterized by cyclonic wrap-up and the other by anticyclonically tilted troughs that thin and form cut-offs. It was found that for the cyclonic life cycles the tropical cyclone moved into the middle latitudes and was absorbed into the large-scale deep surface low pressure system typical of these life cycles. Because of the large difference in horizontal scale between the tropical cyclone and the middle latitude low the TC behaved rather passively, although the presence of the remnant TC PV anomaly lead to enhanced surface winds. For the anticyclonic life-cycles the sensitivity to the initial TC position was strong, with the middle latitude trough moving past the TC vortex in some cases so that no ET took place and in other cases the TC vortex interacted with the small-scale tropopause cut-off or formed a frontal wave and developed into a small scale but rather intense middle latitude low. Singular vectors (SVs) constructed from the adjoint model of the U. S. Naval Operational Global Atmosphere Prediction System for three Atlantic hurricanes in 2004, Ivan, Jeanne and Karl, have been examined to understand interactions between them and a middle latitude trough system (Maue et al. 2006). By optimizing the perturbation energy localized in a small region centered at the 48-hour projected position of a tropical cyclone, the initial SV represents the sensitive region to the final state within the specified region for a specified optimization period. For Hurricane Ivan, the SV analysis reveals the merging of a weak middle latitude trough and Ivan to form a new trough/cut-off low. This new trough system impacted the evolutions of Hurricane Jeanne in subsequent time through the upstream flow of the trough that moves toward Jeanne. The SV associated with Jeanne at a later stage shows that Jeanne influenced the third hurricane Karl though the trough system as Karl went through ET, and became part of the trough. The most striking result is where the final disturbance shows clear signal of Karl when the optimization is confined in the vicinity of Jeanne only. This energy has to come from the trough that is present in the initial sensitivity to Jeanne as there is no signal of Karl in the same initial SVs. This study demonstrates the capability of the SV approach in providing evidence of the complicated interactions between a mid-latitude trough and tropical cyclones. A number of recent studies have used modern models consisting of the atmosphere, ocean waves, sea spray and the upper ocean to evaluate the coupled atmosphere-ocean impacts of surface fluxes, spray evaporation and wave drag on middle latitude cyclones that include systems originating at tropical cyclones (Perrie et al. 2005; Zhang et al. 2006; Ren and Perrie 2006; Ren et al. 2004; Perrie at al. 2004; Zhang and Perrie 2006a; Perrie 2006). These studies focus on the role of air-sea fluxes on storm intensity and development, and related impacts on the structure of the atmospheric boundary layer, and on ocean surface waves and surface fluxes (Zhang and Perrie 2006b). Case studies have included Earl (1998), Daniel (1998), Gustav (2002), and Juan (2003) and two intense winter storms from 2000 and 2002. Results suggest that sea spray tends to intensify storms whereas wave-related drag tends to diminish intensity. Sea-surface temperature depression from the upwelling induced by storm passage represents a further factor that tends to diminish storm strength. The mechanisms by which spray and wave-related drag can influence storm intensity are quite different. When wind speeds are high and sea surface temperatures (SSTs) are warm, spray can significantly increase the surface heat fluxes. By comparison, momentum fluxes related to wave-drag are important over regions of the storm where

323

young, newly generated waves are prevalent, for example during the rapid-development phase of the storm, and decreases in areas where the storm waves reach maturity. The peak flux enhancement values tend to occur during the storm’s intensifying period, when the sea-state is young and rough, and winds are high, and spray droplet production rates are high, into the lower atmosphere. Our studies show that the collective influence of spray and waves on storm intensity depends on their occurrence in the early stages of a storm’s rapid intensification phase, and their spatial distribution with respect to the storm center. Spray evaporation causes the lower part of the atmosphere to experience cooling and moistening. This cooling process increases the air-sea temperature difference, destabilizes the surface layer, and enhances the surface layer turbulence. Thus, the convergence of mass and moisture fluxes from the surface are enhanced, particularly when the local heat flux region is close to the active storm region, as when storms are over warm Gulf Stream waters. This results in upward transport of moisture from both the surface and spray, coupled with latent heat release at the middle tropospheric levels which contributes to warming the mid-troposphere air and lowering the surface pressure. This contribution (of moistening processes) to the thermal-dynamic structure is favorable for storm intensification. By comparison, the influence of wave-drag on storm development is quite different from that of spray. Because of friction-induced kinetic energy dissipation associated with enhanced surface roughness, wave drag can induce anomalous convergence-generated upward motion around the storm center and an attendant dynamic compensation downdraft outside of the storm center, contributing to the downward mixing of upper level dry cool air. This gives a slight increase in static stability in the lower troposphere, thus suppressing convection and reducing storm intensity, and is analogous to the classical Ekman spin-down mechanism. A PV analysis can be used to show that baroclinic processes associated with surface friction induced by wave drag along the warm front also occur, and also can be the dominant mechanism to effect storm development. The inclusion of sea spray can significantly dilute the Ekman pumping effect around storm center, which relegates wave drag effects to secondary importance for storm development. 2.5.3 Downstream Impacts Because the western North Pacific is the region in which most ET processes occur (Jones et al. 2003), it follows that downstream impacts from cyclogeneses in this region be studied. Baroclinic energy conversion in the western and central North Pacific generates a vast amount of kinetic energy that plays a key role in maintaining the storm tracks downstream over the eastern North Pacific, North America, and North Atlantic (Chang and Yu 1999; Orlanski and Sheldon 1995; Nielsen-Gammon and Lefevre 1996, Danielson et al. 2004). Cyclogenesis over the western and central North Pacific is typically triggered by dynamical influences from the surrounding area, including upper-tropospheric wave packets that propagate from East Asia, due to forcing by the extratropical transition (ET) of tropical cyclones, and by Rossby waves initiated by organized tropical convection. The resulting extratropical storms from ET in the western North Pacific basin may have unusually strong downstream effects due to the extraordinary amount of kinetic energy generated during ET. Palmén (1958) estimated that two or three typical ET would provide the entire NH north of 300N with the kinetic energy sufficient to maintain the general circulation against frictional dissipation. Recent papers (Hakim 2003 and Chang 2005) have provided strong statistical evidence that the presence of wave packets on the two Asian waveguides increases the likelihood of the occurrence of deep cyclones over the western and central North Pacific. The northern waveguide crosses through Siberia, while the southern waveguide runs along the subtropical jet across southern Asia (Chang and Yu 1999; Hoskins and Hodges 2002). It has been hypothesized (Chang 2005) that wave packets originating from the northern waveguide trigger cyclogenesis, while wave packets originating from the region of the southern wave guide only enhance, but do not trigger, cyclogenesis. There are also indications that these wave packets are in turn invigorated by the cyclogenesis events, making their

324

impacts further downstream potentially more significant. A case study of the extratropical transition of Hurricane Erin in September 2001 (Roebcke 2003; Roebcke et al. 2004) a sequence of forecasts has been performed using the global model (GME) of the German Weather Service (DWD). Forecasts were initialized each day between the time period 10 to 17 September. Each model run was initialized with European Centre for Medium-Range Weather Forecasts (ECMWF) and DWD analyses. It is suggested that the re-intensification as an extratropical cyclone occurred because Erin moved into a favorable location both from a PV-theta perspective since Erin was located directly to the east of a tropopause depression and from jet streak dynamics since Erin was located between the entry region of a downstream jet streak and the exit region of a smaller-scale upstream jet streak. It is demonstrated that the forecasts are not able to predict the rapid re-intensification phase of Erin. It is suggested that the forecasts' failure can be partly attributed to the model's inability to represent the upstream jet streak well. This appears to be linked to the model's poor representation of Tropical Storm Gabrielle. From a PV-theta perspective and from trajectory calculations it is hypothesized that the outflow of Gabrielle contributed to enhanced baroclinity and a steepened tropopause in the region of the upstream jet streak. As a consequence, the upstream jet streak was enhanced and contributed to Erin's re-intensification. The outflow of Erin is depicted as a raised tropopause downstream of Erin from a PV-theta perspective. Time series of tropopause maps and trajectory calculations suggest that the pronounced ridging downstream of Erin can be attributed to the outflow of Erin. The downstream impact of the outflow extends across the Atlantic basin and influences western Europe. The ridging downstream of Erin is underpredicted in almost all forecasts. Full physics numerical experiments with idealized initial conditions to investigate the impact of a tropical cyclone on the middle latitude flow with focus on the downstream region of an ET-event (Riemer 2006). In this experiment, a TC interacts with the most simplified representation of the middle latitude flow regime: a straight jet. The prominent features of the interaction are a jet streak that forms in the region where the TC outflow impinges on the jet and a ridge-trough couplet on the tropopause. Both features amplify during the interaction. Beneath the left exit region of the jet streak rapid surface cyclogenesis takes place. At upper levels the ridge-trough pattern extends downstream as a wave pattern and initiates a family of cyclones. The upper-level wave pattern can be interpreted as the excitation of a Rossby wave train (RWT) by the ET event and its subsequent propagation downstream. Sensitivity experiments reveal the importance of the atmospheric states in the middle latitudes for the propagation of the RWT. Baroclinic energy conversion and diabatic processes are found to be important for the amplitude of the RWT. The results suggest that the concept of baroclinic downstream development (Orlanski and Sheldon 1995, Nielson-Gammon and Lefevre 1996) is of greater importance than the pure propagation of a (barotropic) RWT. Piecewise PV inversion shows that the cyclonic circulation of the decaying positive PV tower of the ET system is the main contributor to ridge building. The balanced effect of the outflow layer is mainly to intensify the downstream trough. Low-level temperature and PV anomalies have only negligible impact. The effect of the outflow layer on trough formation is confirmed by sensitivity experiments using different cyclone structures. ET events of TC with a less pronounced outflow layer exhibit less meridional orientation of the downstream flow and the primary downstream system tends to a more cyclonic life cycle than in experiments with prominent outflow layers. The effect of the TC structure on the downstream region is greatest one wavelength downstream of ET. The impact on the further propagation of the RWT is more complex and depends amongst other things on moist processes in the environment (Riemer and Jones 2006). A developing baroclinic wave is a more realistic representation of the middle latitudes and we investigate the interaction of a TC with a variety of baroclinic life cycles. Interacting with a mature

325

system the ET event shows only little and localised impact on the upper-level middle latitude flow pattern. Interaction with a developing wave from a localised upper-level perturbation shows significant enhancement of the ongoing baroclinic development and a downstream impact comparable to the straight jet case. These examples show that there are situations in which the middle latitude flow is more conducive for an impact of the ET system. These situations seem to be linked to the maturity of the baroclinic system (Riemer 2006). Research at the University of Washington (R. Torn and G. Hakim) examines the sensitivity to the phasing relationship between the tropical cyclone and extratropical waves, by using an ensemble Kalman filter (EnKF) for select western Pacific ET events. Previous research has shown that small changes to the position of the TC or mid-latitude waves can significantly alter the ET forecast and thus the downstream state (e.g. Browning et al. 2000, Klein et al. 2002). One advantage of this technique is that is provides an analysis ensemble of equally likely states, and thus a way to objectively measure relationships between the tropical cyclone and extratropical circulations. Preliminary results show that the downstream impact of ET is proportional to how much the TC reintensifies as a baroclinic system. For the case of Typhoon Tokage (2004), those ensemble members that include a deepening baroclinic storm produced a downstream Rossby wave, while those ensemble members that include a weakening TC have little downstream impact. Additional research is needed to understand the physical processes that lead to the aforementioned rapid error growth in the downstream state. 2.5.4 Summary and recommendations for future research directions A continuing science issue is to understand the origins of the varying types of extratropical transition, including being able to identify the physical mechanisms that allow a subset of these storms to reintensify explosively. Another crucial science issue is to what extent differences in mean environmental conditions across the various ocean basins can contribute to varying types of extratropical transition. Another science issue is determining what is the collective impact on the general circulation of the atmosphere of the periodic insertion of upper-level warm pools into middle and high latitudes in conjunction with the ET process. As pointed out earlier, this has impact on predictability. Factors associated with reduced forecast skill include parameterization of convection, and oceanic surface fluxes. A continuing operational issue is the need to understand the dynamical processes control the distribution and amount of precipitation relative to the tracks of landfalling and transitioning tropical cyclones. Work is underway on transferring research knowledge on this topic to operations through collaborative efforts with Wes Junker at NCEP/HPC. Future work should also address the sensitivity of the downstream state to the upstream state and the tropical cyclone during ET events. Future investigation of the relative roles of fluxes and momentum transports on ET strength is needed. Bibliography Abraham, J., W. Strapp., C. Fogarty, and M. Wolde, 2004: Extratropical transition of Hurricane Michael: an aircraft investigation. Bull. Amer. Meteor. Soc., 85, 1323–1339. Atallah E., et al. 2006: Precipitation distribution associated with landfalling tropical cyclones over the eastern United States, Mon. Wea. Rev., conditionally accepted. Browning, K. A., A. J. Thorpe, A. Montani, D. Parsons, M. Griffiths, P. Panagi, and E. M. Dicks, 2000: Interactions of tropopause depressions with an ex-tropical cyclone and sensitivity of forecasts to

326

analysis errors. Mon. Wea. Rev., 128, 2734-2755. Chang, E. K. M., 2005: The impact of wave packets propagating across Asia on Pacific cyclone Development. Mon. Wea. Rev., 133, 1998-2015. Chang, E. K. M., and D. B. Yu, 1999: Characteristics of wave packets in the upper troposphere. Part I: Northern Hemisphere winter. J. Atmos. Sci., 56, 1708-1728. Danielson, R. E., J. R. Gyakum, and D. N. Straub, 2004: Downstream baroclinic development among forty-one cold-season eastern North Pacific cyclones. Atmosphere-Ocean, 42, 235-250. Dickinson, M. J., K. Corbosiero and L. F. Bosart: 2006: The extratropical transitions of eastern Pacific Hurricane Lester (1992) and Atlantic Hurricane Andrew (1992): A comparison. Mon. Wea. Rev., in preparation. Fogarty, C. T., 2006: The extratropical transition of Tropical Storm Ophelia (2005): Summary of forecasts and meteorological observations. Proceedings from the 40th annual Congress of the Canadian Meteorological and Oceanographic Society. 29 May – 01 June, 2006, Toronto, Ontario, Canada. http://www.cmos2006.candac.ca/ Fogarty, C. T., R. J. Greatbatch, and H. Ritchie, 2006a: The role of anomalously warm sea surface temperatures on the intensity of Hurricane Juan (2003) during its approach to Nova Scotia. Mon. Wea. Rev.,. 134, 1484-1504. Fogarty, C. T., R. J. Greatbatch, and H. Ritchie, 2006b: A numerical modeling study of the extratropical transition of Hurricane Michael (2000). Wea. Forecasting, submitted. Hakim, G. J., 2003: Developing wave packets in the North Pacific Storm Track. Mon. Wea. Rev., 131, 2824-2837. Harr, P.A., and R. L. Elsberry, 2000: Extratropical transition of tropical cyclones over the western North Pacific. Part I: Evolution of structural characteristics during the transition process. Mon. Wea. Rev., 128, 2613-2633. Harr, P. A., D. Anwender, and S. C. Jones, 2004: Predictability associated with the extratropical transition of tropical cyclones as defined by operational ensemble prediction systems. Preprints, 26th Conference on Hurricanes and Tropical Meteorology, American Meteorological Society, Boston, MA 683-684. Hoskins, B. J., and K. I. Hodges, 2002: New perspectives on the Northern Hemisphere Winter Storm Tracks. J. Atmos. Sci., 59, 1041-1061. Jones, S., P.A. Harr, J. Abraham, L.F. Bosart, P.J. Bowyer, J.L. Evans, D.E. Hanley, B.N. Hanstrum, R.E. Hart, F. Lalaurette, M.R. Sinclair, R.K. Smith, and C. Thorncroft, 2003: The extratropical transition of tropical cyclones: Forecast challenges, current understanding and future directions. Wea. Forecasting, 18, 1052-1092. Kelsey, and L. F. Bosart (2006): The extratropical transition and explosive reintensification of Supertyphoon Dale (1996). Mon. Wea. Rev., in preparation. Klein, P. M., P. A. Harr, and R. L. Elsberry, 2002: Extratropical transition of western North Pacific tropical cyclones: Midlatitude and tropical cyclone contributions to reintensification. Mon. Wea. Rev., 130, 2240-2259.

327

Maue et al. 2006: Hurricanes Ivan, Jeanne, Karl (2004) and Mid-latitude Trough Interactions. 27th Conference on Hurricane and Tropical Meteorology, Monterey, 24-28 April 2006. Moore, R. W., and M. T. Montgomery, 2005: Analysis of an idealized three-dimensionaldiabatic Rossby vortex: A coherent structure of the moist baroclinic atmosphere. J. Atmos. Sci., 62, 2703-2725. McTaggart-Cowan, R, L. F. Bosart, J. R. Gyakum, and E. H. Atallah, 2006a: Evolution and global impacts of a diabatically-generated warm pool: Hurricane Katrina (2005). Mon. Wea. Rev., submitted. McTaggart-Cowan, R., E. Atallah, J. R. Gyakum, and L. F. Bosart, 2006b: Hurricane Juan (2003). Part I: A diagnostic lifecycle study. Mon. Wea. Rev., 134, 1725-1747. McTaggart-Cowan, R., E. Atallah, J. R. Gyakum, and L. F. Bosart, 2006c: Hurricane Juan (2003). Part II: Forecasting and numerical simulation. Mon. Wea. Rev., 134, 1748-1771. McTaggart-Cowan, R., L. F. Bosart, C. A. Davis, E. H. Atallah, J. R. Gyakum, and K. A. Emanuel, 2006d: Analysis of Hurricane Caterina (2004). Mon. Wea. Rev., in press. Nielsen-Gammon, J. W., and R. J. Lefevre, 1996: Piecewise tendency diagnosis of dynamical processes governing the development of an upper-tropospheric mobile trough. J. Atmos. Sci., 53, 3120-3142. Orlanski, I., and J. P. Sheldon, 1995: Stages in the energetics of baroclinic systems. Tellus, 47A, 605-628. Palmén, E., 1958: Vertical circulation and release of kinetic energy during the development of hurricane Hazel into an extratropical storm. Tellus, 10, 1-23. Perrie, W., 2006: Editor of “Atmosphere-Ocean Interactions. Vol. 2”. Wessex Institute of Technology. 224 pp. Perrie, W., X. Ren, W. Zhang, and Z. Long, 2004: Simulation of extatropical Hurricane Gustav using a coupled atmosphere-ocean-sea spray model. Geophys. Res. Lett., 31, L03110, doi:1029/2003GL018571. Perrie, W., E. Andreas, W. Zhang, W. Li, E. L, J. R. Gyakum, and R. McTaggart-Cowan, 2005: Impact of sea spray on rapidly intensifying cyclones at midlatitudes. J. Atmos. Sci., 62, 1867-1883. Ren, X., W. Perrie, Z. Long, J. Gyakum, and R. McTaggart-Cowan, 2004: On the atmosphere-ocean coupled dynamics of cyclones in midlatitudes. Mon. Wea. Rev., 132, 2432-2451. Ren, X., and W. Perrie, 2006: Air –sea interaction of typhoon Sinlaku (2002) simulated by the Canadian MC2. Advances in Atmospheric Science. in press. Riemer, M., 2006: The impact of extratropical transition on the downstream flow: idealised modelling study, Preprints of the 27th Conference on Hurricanes and Tropical Meteorology, Monterey, 24-28 April 2006. Riemer, M., and S. C. Jones, 2006: The impact of extratropical transition on the downstream flow: idealised modelling study with a straight jet. Q.J.R.M.S, in preparation. Roebcke, M., 2003: An investigation of the extratropical transition of Hurricane Erin using the Global Model (GME) of the Deutscher Wetterdienst. Diplom thesis, University of Munich, 67pp. Available from

328

sarah.jones@imk.uka.de. Roebcke, M., S.C. Jones, and D. Majewski, 2004: The extratropical transition of Hurricane Erin (2001): a potential vorticity perspective. Meteorol. Z., 13, 511-526. Sardeshmukh, P. D., and B. J. Hoskins, 1988: The generation of global rotational flow by steady idealized tropical divergence. J. Atmos. Sci., 45, 1228-1251. Weindl, H., 2004: Numerical experiments on the interaction of a hurricane-like vortex with a baroclinic wave. AMS preprints, 26th Conference on Hurricanes and Tropical Meteorology. Miami, Florida, 3-7 May 2004. Wolde, M., D. Marcotte, J. Jordan, J. Aitken, J. Abraham, and J. W. Strapp, 2001: The First Canadian Experience with Research Flight Operations in Hurricane Extratropical Transition. Canadian Aeronautics and Space Journal, Vol. 47, No. 3, 179-189. Zhang, W., and W. Perrie, 2006a. Extratropical Hurricane Juan: Structure and Maintenance. Mon. Wea. Rev., submitted. Zhang, W., and W. Perrie, 2006b: Impacts of storm-induced sea spray and wave drag on ocean waves. Ocean Modelling, submitted. Zhang, W., W. Perrie, and W. Li, 2006: Impacts of waves and sea spray on midlatitude storm structure and intensity. Mon. Wea. Rev., in press.

329

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES Topic 2a : The Catarina Phenomenon Rapporteur: Pedro L. Silva Dias University of São Paulo Institute of Astronomy and Geophysics 05508-900 Sao Paulo – BRAZIL E-mail: pldsdias@master.iag.usp.br Fax: 55-11-3091-4714 Working Group: Manoel Gan, John L. Beven, Alexandre Pezza, Greg Holland, Augusto Pereira, Ron McTaggart-Cowan, Francisco de Assis Diniz, Marcelo Seluchi, Hugo J. Braga 2.a.1 Introduction Sea Surface Temperatures (SSTs) warmer than 26.5oC and Environmental Vertical Wind Shear (EVWS, defined as the magnitude of the difference between the 200 and 850 hPa vector winds) lower than 8 m/s offer ideal conditions for Tropical Cyclone (TC) development. Thus it has been accepted that hurricanes could not form over the South Atlantic Ocean due to the very intense climatological vertical wind shear and not sufficiently warm SST. However, on March 28, 2004 the cyclone Catarina hit the southern region of Brazil (State of Santa Catarina). This was the first documented time when a system reaching a category I hurricane strength made landfall anywhere in the South Atlantic basin. This is not to say that a phenomenon like the Catarina Hurricane had not existed in the past, but there is very strong evidence that at least during the satellite era this is unprecedented.

A few important questions arise after March 2004. First, what Catarina really was and how should we refer to it? Second, was Catarina a result of natural climate variability only, or could it also be related to climate change due to anthropogenic influences? 2.a.2 Synoptic History Observations of the Catarina event include satellite-based Dvorak and Hebert-Poteat technique intensity estimates from the Tropical Analysis and Forecast Branch (TAFB) of the Tropical Prediction Center (TPC) in Miami, Florida, the Satellite Analysis Branch (SAB) in Washington, DC, and AFWA. Microwave satellite imagery from NOAA polar-orbiting satellites, the NASA Tropical Rainfall Measuring Mission (TRMM), the NASA QuikSCAT, the NASA Aqua, and Defense Meteorological Satellite Program (DMSP) satellites were also useful in tracking the cyclone. Operational analysis of the global forecasting centers also indicate the structure of the system although their generally rather coarse resolution impairs a detailed analysis of the structure of the system. The origin of the phenomena goes back to about 8 days before the cyclone hit the coast. An intense cold front was followed by the formation of a low pressure system at about 22oS. This development was a typical baroclinic cyclogenesis caused by an upper-level trough (Fig. 2.a.1) interacting with a surface frontal system, with satellite imagery showing features typical of that mode of development (Fig.2.a.2). The Catarina Cyclone trajectory was derived from the University of Melbourne automatic tracking algorithm showing the central locations every 06 hours (Fig. 2.a.3), based on the high resolution ECMWF analysis (0.5o), and the maximum SSTs (oC) for the period between 20 and 28

330

March. On 23 March, the upper-level trough was cutting off from the westerlies to form an upper-level low (Fig. 2.a.1). This resulted in the surface low turning west-northwestward, characterizing a comma cloud system (Fig. 2.a.2d and Fig. 2.a.3) embedded in a baroclinic environment, a common feature in this region during autumn and spring according to Bonatti and Rao (1987). Early attempts to classify this weather system also included a polar low event because in the early stages of development it produced a typical comma cloud pattern, sometimes observed, for instance, over the Mediterranean Sea (Reale and Atlas, 2001) where they sometimes resemble tropical cyclones with the eye structure and relatively shallow convection compared to the tropical standards. However, the NASA QuikSCAT instrument indicated up to 18 ms-1 winds associated with the surface low (Figure not shown), and convection first developed near the surface low center. The system was quite strong compared to other comma systems observed in the region. Central convection increased on 24 March and the comma cloud shape became very well defined (Fig. 2.a.2f). Operational services looked more closely at the situation because there were several indications that it was an anomalous system. Comma clouds in this area usually slowly displace east/southeastward. However, this particular system was moving north-westward at this time (Fig 2.a.3). The U. S. Air Force Weather Agency (AFWA) in Omaha, Nebraska made the first subtropical cyclone intensity estimates using the Hebert-Poteat (1975) technique. The AFWA estimated that the cyclone became a subtropical storm around 00 UTC 25 March about 900 km from the coast at 28oS (Fig. 2.a.2g) although at that time the system still had the typical characteristics of the comma systems, according to the Center for Weather Forecasting and Climate Research (CPTEC) bulletin. Catarina turned westward later that day with development continuing. The convection wrapped around a formative eye near 00 UTC 26 March (Fig. 2.a.2i), with the eye becoming embedded in a convective overcast by 15 UTC that day (Fig. 2.a.2j). The westward displacement of Catarina indicates that it followed an optimum path to avoid strong vertical wind shear. Northward and southward of the vortex wind shear would have destroyed its vertical structure and the system would dissipate (Nakano and Nakajima, 2004). Sea surface temperatures, on the other hand, were higher on the coast where the warm current of Brazil developed a tongue of warm waters. The classification of the vertical structure of the system, based on the Hart (2003) system (http://moe.met.fsu.edu/cyclonephase/) with the 1o resolution NCEP/AVN analysis and the AMSU soundings processed at CIRA are shown in Figure 2.a.4. The vertical cross section of the NCEP/AVN high resolution operational analysis (0.5o) is shown in Figure 2.a.5. Figure 2.a.5 indicates the intrusion of warm air at the top of the troposphere on 23 March and that the low level warm core is very narrow. Figure 2.a.5 also shows the diabatic heating derived from the NCEP high resolution analysis. Maximum heating at the earlier stages (Fig. 5c) seems deeper than at the final stage (Fig. 2.a.5d) when convective heating reached about 25-30oCday-1 at a 800 hPa. Coarser operational analysis failed to properly identify the inner warm core as can be seen in Figure 2.a.4a (based on the NCEP/AVN 1o resolution analysis). Similar analysis based on other operational analysis (e.g., the Canadian Meteorological Service) also failed to identify the inner warm structure of the system (figure not shown). Figure 2.a.6 shows the AMSU-derived azimuthally-averaged temperature anomalies for the pre-Catarina low near 21 UTC 23 March (courtesy of the Cooperative Institute for Research in the Atmosphere/CIRA at Colorado State University). Clearly, the warm core is restricted to the upper troposphere/lower stratosphere. Figure 2.a.6 shows a cold-core cyclone in the middle levels (6-7 km), with a large and strong warm-core signature at 14-16 km. This is consistent with a baroclinic cyclone, with cold air in the middle levels and warm anomalies at high levels due to a lowered tropopause associated with the upper-level trough (Fig. 2.a.1). At this time, Hart classification applied to the Advanced Microwave Sounding Unit (AMSU) sounding (Fig. 2.a.4b) classified the system as moderately warm core because of the analysis is restricted to the 900-600hPa and 600-300hPa layers, where the temperature anomalies were not well defined in the early stages. The AMSU soundings indicated significant warming of the lower and middle levels on 26 March compared to the

331

structure shown in Figure 2.a.6. It also shows a smaller, but more intense, warmer core now concentrated over the center of the cyclone. This anomaly is at 10-12 km, or about the height of the tallest observed convective tops. These signatures are more characteristic of that of a tropical cyclone, and intensity estimates (Brueske and Velden 2003) based on the AMSU-measured warm core indicated Catarina was of hurricane strength by March 26. Based on this and high resolution analyses of temperatures from the AMSU and the NCEP/AVN 100 km resolution analysis, it is estimated that Catarina completed its transformation to a narrow warm core system around 12 UTC 26 March. Only high resolution analysis were able to properly identify the inner structure of the core. Figure 2.a.4b suggests that a rapid transition to moderately warm core took place on 23 March. Bonatti and Rao (1987) indicates that comma cloud systems in this region tend to follow a similar evolution of the inner thermal structure. However, the Catarina Cyclone continued to evolve into the tropical transition (TT), contrary to the typical situation of the comma systems. While Catarina may be unique in the South Atlantic, this type of evolution is often seen in the North Atlantic, as typified by Hurricane Epsilon (Franklin 2006). Catarina continued to move westward on 26 March between the cut-off low to the north and a mid/upper-level anticyclone to the south (Fig. 2.a.3). This was followed by a west-southwestward motion the next day. The cyclone maintained a well-defined eye with Dvorak (1984) tropical cyclone intensity estimates of 38-45 ms-1 during this time. It turned west-southwestward on 27 March, and this motion brought the center to the coast of Brazil near 29oS at the Santa Catarina coast at about 05 UTC 28 March (Fig. 2.a.2 and Fig. 2.a.3). As the warm core formed, Catarina also developed some anticyclonic outflow, which is another aspect of a ‘classic’ tropical cyclone. Figure 2.a.7 shows upper-level satellite-derived winds produced by the Cooperative Institute for Meteorological Satellite Studies (CIMSS) at the University of Wisconsin for 1800 UTC 27 March. Note the wind barbs in the cyan (greenish-blue) color to the south and southeast of Catarina representing winds between 100-250 hPa. These winds are radially outward from the center south of the cyclone, then turn anticyclonically to become westerlies to the southeast of Catarina Satellite imagery showed cooling of the convective cloud tops in the last few hours before landfall on 28 March, suggesting that Catarina was strengthening as it reached the coast. This may have been due to passage over the tongue of warmer sea surface temperatures just offshore of the landfall area (Figure 2.a.3) associated with the Brazil current. Catarina continued generally west-northwestward after landfall, moving into the higher terrain of southeastern Brazil where it dissipated late on 28 March (Fig. 2.a.2). An analysis of the precipitation associated with Catarina was performed by Pereira and Lima (2006). They show the satellite daily maximum rainfall estimates between 23 and 29 March obtain with CMORPH (Joyce et al., 2004). The daily maximum precipitation associated with Catarina was fairly constant and above 100 mm day-1 except on 25 and 29 March. During and after Catarina, questions were raised regarding how convective the clouds surrounding the eye actually were. Convective cloud tops over the Amazon areas of Brazil often reach heights of 20 km, while those in Catarina were much lower in the 6-12 km range as shown by convention satellite imagery and radar cross sections from the TRMM satellite. Figure 2.a.5 also suggests that the maximum convective heating level was significantly lower than the typical values found in the tropical sector of South America (although at some stages, such as in Fig. 2.a.5c when it reached 200 hPa with peak values between 600 and 500 hPa). High resolution imagery at landfall indicates the development of an intense convective system in the center of the eye and on the southeastward corner of the eye wall (Fig. 2.a.8). These convective cells were detected by the automatic tracking system available at CPTEC (Machado and Laurent 2004). The cell at the eye center may have been triggered by the interaction with the abrupt raise in the local topography (Figure 2.a.3). There are several indications

332

of tornadoes and microbursts associated with the intense convective cells near at the center of the eye and at the eye wall as reported by observers and based on the damage produced by the storm. Additionally, a TRMM radar cross section on 27 March (Fig. 2.a.9) also shows convective towers extending above the ‘bright band’ melting/freezing level, particularly to the left of the eye on the figure. These data show the presence of convective clouds in the cyclone. The map of lighting strikes (figure not shown) indicates the absence of electrical activity associated with Catarina particularly in the eye wall, though convective cells had a vertical structure that indicated the presence of ice crystals (Pereira and Lima, 2006). It might be related to the kind of cloud condensation nuclei (CCN) available over the ocean (salt). Some electrical activity was observed over the continent in association to a convective band to the Northwest of Catarina. It is likely that the relatively low convective heights are due to the lowered tropopause associated with the upper level trough (Fig. 2.a.1). Such suppression of the cloud tops has been seen in similar North Atlantic storms (e. g., Hurricane Epsilon). Epsilon also formed and spent much of its life over SSTs of 23C. The most likely reason is that the cold environment of the parent baroclinic system creates sufficient vertical instability to drive convection even over the relatively cool water. Emanuel (1999) indicates that these SSTs were sufficient to support a storm of hurricane strength given the relatively cool air temperatures present. The only ground truth for the intensity of Catarina was a few surface observations in the landfall area. Figure 10 shows hourly data at São Bento (28o 36’S, 49o 33’W, 135 m), just north of eye at landfall) between the 26th and the 28th of March 2004 (local time) of station pressure (elevation 135 m) and 10 m height wind speed (Pezza and Simmonds, 2006). This meteorological station was the closest available to the region where the eye made landfall, being about 40 km directly to the north of the eye at that time. Accounts from the landfall area described a perfect eye passage (Reynaldo Haas, personal communication), with the winds blowing very strongly just before a period of calm, followed immediately by a second round of strong winds that gradually died away. At Torres, about 60 km from the center of the storm, pressure drop of the order of 10 hPa in 6 hr was observed at a station of the Brazilian National Institute of Meteorology (figure not shown). Assuming axisymmetry and the observed propagation speed of the eye and using the gradient wind approximation, winds of the order of 25-30 m/s are compatible with the observed pressure drop. 2.a.3 Role of Blocking and Vertical Shear In a recent study, McTaggart-Cowan et al. (2006) investigate the initiation and development of Hurricane Catarina from a diagnostic perspective. They show that an anomalously persistent dipole blocking episode over the western South Atlantic Ocean (Fig. 2.a.11) is responsible for reducing the climatologically high shear values in the area to a level suited for tropical cyclogenesis that follows a tropical transition (TT) pathway triggered by a weak extratropical cyclone (WEC) (Davis and Bosart 2004). The dipole block is maintained by repeated injections of high (low) potential vorticity air from the central South Pacific Ocean into the ridge (trough) component of the blocking structure over a period of 13 days leading up to and during Catarina (Fig. 2.a.1). et al. (2006) performed a theoretical analysis of the group velocity of rotational waves in the environment associated with Catarina. They show that the relative vorticity stretching and secondly in importance is the planetary vorticity stretching. The planetary vorticity advection does not contribute in this case. The dipole blocking pattern is shown to play three distinctly beneficial roles in the development and maintenance of Catarina as a tropical vortex. Initially, the cooler upper level (defined in the study as the dynamic tropopause, 2 PVU surface) air in the trough component of the block reduces the bulk column stability over a broad area off the Brazilian coast. This is reflected by potential intensity values (Bister and Emanuel 2002) that are well above climatological norms for the region, despite the relatively (and anomalously) cool sea surface temperatures that do not exceed 25oC during Catarina’s lifecycle (Fig. 2.a.3). Furthermore, the weak, anomalous easterly flow at upper levels allows convection to organize without the deleterious impact of strong vertical shear. These combined destabilizing and

333

shear reduction effects are sufficient to allow a precursor disturbance moving southeastward beneath the block from the South American continent to begin a TT process. Pezza and Simmonds (2005) recently explored some aspects of the large scale circulation which may be related to the development of Catarina. The Environmental Vertical Wind Shear (EVWS) is defined as the magnitude of the difference between the 200 and 850 hPa vector winds (in m/s). A shear index was defined as the average of the EVWS between 35 and 60oW along the 30oS latitude. This domain is representative of the midlatitude environment in the environs of Catarina’s track. Figure 12 shows the anomaly of the EVWS magnitude averaged for the period 23rd – 28th March for the South American sector. A pronounced negative anomalous shear region between 25 and 40oS and 35 and 60oW is observed lying just to the south of the cyclone track, with a shear anomaly of -20 m/s next to the place where the system made landfall. The negative 15 m/s shear anomaly roughly corresponds to mean values below the ideal threshold of 8 m/s. The vortex itself may have exerted only a very limited influence in the EVWS given the small scale of the Catarina and the fact that the anomalies presented a well defined large scale pattern and were present before the TT started, with negative anomalies prevailing in all longitudes around 30oS suggesting a blocking pattern to the south.

A blocking-like index (B) was defined as the average geopotential anomaly in the area between 47.5 and 55oS and 20 and 60oW by Pezza and Simmonds (2005). From a dynamic point of view it is expected that when this index is high the westerlies will be weaker than normal at midlatitudes in the South American sector, therefore being associated with less large scale baroclinicity (also depending on the static stability) and physically consistent with the TT (Hart, 2003; Davis and Bosart, 2003). However, the association between the blocking index and the shear is not direct because the former pertains to conditions at higher latitudes. The temporal series were calculated for the 1979-2004 period, when the NCEP/DOE reanalysis II dataset offers a reliable climatology to put Catarina in perspective with the natural variability in the last 25 years (Figure 2.a.13). For the unfiltered data, the minimum EVWS index during Catarina was 7.0 m/s (only +1.8 m/s for the u-component) and the maximum B index corresponded to +181 geopotential meters. Further indicating the extreme and large scale nature of the circulation anomalies leading up to the event, Pezza and Simmonds (2005) have also found that the 5-point average of the B index during Catarina was exceeded for only 0.62% of the record of all Marchs 1979-2004. This shows that from a climatological point of view the blocking-like pattern at mid-to-high latitudes was very intense. In addition the low EVWS phase which started seven days after the blocking peak was exceptionally long, with almost five consecutive days with below 12 m/s EVWS index during the whole time (06 hourly data) in a region subjected to high climatological shears (25.7 ± 8.8 m/s for the EVWS index). Only 0.40% of the total unfiltered six-hourly sample exhibited an EVWS index below the minimum during Catarina (0.25% for the filtered data presented in Fig 13), but such condition was only seen during March 1993 when the blocking index did not show any significant positive anomaly. Although the temporal variability in both series is high, the combination of a B index higher than Catarina followed by a wind shear lower than Catarina is not found anywhere else in the record, indicating unprecedented conditions for the whole 1979-2004 period. It’s physically consistent to expect that these very anomalous large scale conditions favored the occurrence of the TT, generating the sufficiently low EVWS in the period of the year when the climatological SST is maximum and relatively close to the hurricane threshold, and in the present case when a previously high relative vorticity environment was present over the surrounding area given by the passage of a cold front.

2.a.4 Large-Scale Perspective The first 4 months of 2004 were characterized by significant anomalies in the global circulation. In

334

particular, the intraseasonal variability was particularly strong. Anomalous strong upward sensible and latent heat transfer off the southeastern coast of Brazil was also observed since January (Figures not shown). However, SST was slightly colder than normal. Thus, it is clear that the air temperature at the surface was rather cold. In fact, January 2004 was one of the coldest in 25 years due to the prevailing SE flow associated with a South Atlantic Convergence Zone to the north of its normal position as indicated in Figure 2.a.14 (OLR global anomalies). January 2005 was also exceptionally rainy in NE Brazil as a result of the northward migration of the South Atlantic Convergence Zone - SACZ (Climanalise, February 2004). Table 2.a.1 shows the number of cases of cyclogenesis off the SE coast of Brazil based on the NCEP reanalysis. The minimum and maximum number observed in the climatological analysis of Gan and Rao (1991) is shown in the last column. It is clear that the first three months of 2004 were quite active in terms of cyclogenesis. Thus, in spite of the relatively cold water, the air was anomalous cold and the heat and moisture transfer from the ocean to the atmosphere were stronger than normal (as seen in Pereira and Lima, 2006). Table 2.a.1 Number of observed cyclogenesis in 2004 according to the NCEP reanalysis of the SE coast of Brazil and climatological occurrences (minimum and maximum, respectively , based on Gan and Rao (1991). Figure 2.a.15 shows the time series and the real part of the wavelet spectra of the : Madden Julian Oscillation (Fig. 2.a.15a), Southern Oscillation (Fig. 2.a.15b) and Antarctic Oscillation (Fig. 15c). They all show significant activity in the intraseasonal band during the February-April period. Large amplitude of the intraseasonal activity is also observed in the PNA, PSA and NAO patterns (figures not shown). The SST temperature in the South Atlantic near 32oS also indicated exceptionally high amplitude of the intraseasonal oscillation thus reflecting changes in the atmospheric and oceanic circulation. Thus, it is quite clear that anomalous large scale circulation was observed well before the Catarina event in association with an exceptionally strong intraseasonal oscillation. 2.a.5 Downscaling Experiments Numerical simulations with the mesoscale model RAMS and the Brazilian version BRAMS (www.cptec.inpe.br/brams) were performed to identify the impacts of horizontal resolution, initial condition, convective parameterization, Atlantic SST during the period 23-28 March 2004. A series of extended conference papers were published by the Brazilian Meteorological Society (www.sbmet.org.br) in 2004 (Silva Dias et al. 2004; Menezes and Silva Dias, 2004; Silva et al. 2004; Gevaerd et al. 2004). Most of these experiments were based on the NCEP operational 1o resolution analysis that was available at the time. The mesoscale model was provided with the NCEP analysis as initial and boundary conditions. The higher resolution played a significant role in improving the description of the intensity and the trajectory of the system. Better results were obtained with resolution at least of the order of 8km. However, most of these simulations did not reproduce the observed

8,0 8 March

7,2 10 February

7,3 6 January

Gan & Rao

(1991)

Observed 2004

(NCEP)

335

intensity of the low pressure as it approached land (pressure of the order of 1000 hPa instead of approximately 980 hPa). Figure 2.a.16 shows the predicted precipitation with a 8km resolution BRAMS simulation. The model able to reproduce realistic features of the rain bands and the precipitation rates are compatible with the TRMM derived information (Pereira and Lima, 2006). However, the 10m winds are much weaker than the NASA QuikSCAT estimates. Changes in the convective parameterization were also not successful in improving the lowest pressure and wind intensity at the center of the storm. Higher resolution SST improved the intensification of the system as it approached the coast. However, the high resolution SST data has a small cold bias with respect to the course resolution data. Changes of the order of 1 to 2o C in the SST improved the simulation but the low pressure was still well above the observed value. The high resolution simulations consistently predicted higher cloud tops than observed and therefore the vertical gradient of the diabatic heating close to the surface was relatively small. This is a required condition for large pressure drop at the surface when diabatic heating plays a significant role in the development of deep storms. A set of experiments on the sensitivity of the central pressure on perturbations in the vertical shape of the diabatic heating was recently performed at the University of São Paulo and the results presented by Silva Dias et al (2005) at an special event organized by the Brazilian Meteorological Society in association with the National Institute of Meteorology (INMET) and the Center for Weather Forecasting and Climate Research (CPTEC) (Workshop on the Catarina Event, June 2005, http://www.sbmet.org.br/publicacoes/informativo/2005_07/index.html). Sensitivity to domain size was also explored. Figure 2.a.17 shows the minimum pressure at the center of the cyclone (2.a.17a), maximum 10m wind intensity (2.a.17b) and perturbed trajectories (2.a.17c). It is clear that the vertical shape of the heating profile has the largest impact. In fact, in some cases the cyclone trajectory veers to the north, towards warmer waters, eventually reaching the Rio de Janeiro coast. The best trajectories and low pressure simulation were attained with moderate changes in the heating shape. Domain size did not have a significant impact in the evolution of the cyclone. The change in the heating profile is equivalent to the procedure adopted by the GFDL model (Bender and Marchock, 2004) forecasts of hurricanes. The GFDL model was able to reproduce many important features of the system, including its path, with great accuracy (not shown). This specific model initialization is based on the creation of a synthetically created vortex based on parameters provided to the model that places a warm anomaly in the eye of the vortex. This very interesting and “unorthodox” modeling approach and the dynamical understanding of how to drop the low pressure through latent heating can be applied to probable future events as well as in any other model to spin up the system and explore possible scenarios of the future evolution. 2.a.6 Previous Cases As with any ‘first ever’ event, the question arises as to whether storms similar to Catarina have occurred before. Prior to 2004, the only known tropical cyclone in the South Atlantic was a system of tropical depression- or perhaps minimal tropical-storm strength in April 1991 (McAdie and Rappaport 1991) in the eastern part of the basin. An older example was a system in January 1970, where satellite imagery shows a cyclone with an eye well behind a cold front apparently embedded in cold-air clouds. The tropical nature (or lack thereof) of this system requires more investigation. If possible, a concerted effort should be made to track down similar system in the past using satellite imagery and synoptic data. However, such a search will likely be hindered by the lack of geostationary satellite coverage of the South Atlantic prior to 1966. However, there were two other systems in 2004 that attracted interest – a possible tropical storm near 15S 37W on 18-21 January and a possible hybrid cyclone off on southeastern Brazil on 15-16 March. On 15 May 2004 another well defined eye like structure formed off the southern coast of Brazil (figure not shown). But the system was weaker and rapidly displaced southeastward. It is quite clear that the large-scale conditions were more conducive than usual to possible subtropical or tropical transitions in

336

this area in 2004. Since Catarina, other examples of possible tropical or subtropical cyclones have been noted in real time or found in the past. One case at least somewhat similar to Catarina occurred off of southeastern Brazil on 22-23 February 2006, when a baroclinic cyclone briefly developed an eye and impressive banding in radar data. Questions remain as to how tropical this system was, as it never fully separated from the westerlies or from its parent baroclinic zone. Another cyclone with warm core structure formed along the South Atlantic Zone between 11 and 17 March 2006. This particular system followed a southward trajectory but numerical guidance often indicated an westward migration. McTaggart-Cowan et al. (2006) focuses on other long-lived dipole blocking episodes in the western South Atlantic Ocean (six such episodes are found in the NCEP/NCAR Reanalysis for 1971-2001). The authors find that these events are frequently correlated with the development of severe weather. Of the six events identified in the analog study, two contained wave-like patterns rather than true blocks, two were too weak to halt the progression of systems towards the southeast, one resulted in the development of a strong equivalent barotropic vortex, and one led to a subtropical development. Combining these findings with their analysis of Hurricane Catarina’s development, the authors suggest that tropical cyclogenesis in the area is strongly modulated by the presence of intense and persistent dipole blocking events. 2.a.7 Casualty and Damage Statistics According to the State of Santa Catarina Civil Defense (information available at the site www., in an area with approximately 400,000 inhabitants and 154,000 buildings, 23 cities were severely damaged. Over 33,000 people lost their homes and 40,000 buildings were severely damaged. Roof damage in some cities reached about 95% of the houses. Four persons lost their lives and 7 people disappeared in small boats along the coast. Agriculture damage reached about US$40 million mostly in rice fields (25%), corn (90%) and banana (70%), with 25, 90 and 70%, respectively, production loss in the area affected by the storm. A large number of industries were affected (approximately 1800) and their activities were interrupted for up to a month. Approximately 8600 people lost their jobs as a result of the damages in the industrial and commercial sectors. Roads were blocked and had to be repaved. The State of Rio Grande do Sul, just to the south of Santa Catarina, was also hit by severe winds. About 31,000 people were affected in Rio Grande do Sul along the northern coast and 4500 houses were damaged. A total of about 80% of the schools interrupted classes for up to 15 days. 2.a.8 Forecast and Warning Discussion

Due to the scarcity of tropical cyclones in the South Atlantic, there is no formal World Meteorological Organization (WMO) tropical cyclone program for the area and no formal warning responsibilities. Thus, when Catarina formed, the various warning agencies involved had to organize a response rather quickly.

The formation of Catarina was first formally noted by the satellite analysts at AFWA. They notified the National Hurricane Center (NHC) of the storm’s existence on 25 March, as well as informing other parts of the tropical cyclone community. The NHC, in turn, contacted the Marine Meteorological Service of the Brazilian Navy Hydrographic Service via e-mail and telephone, as they were the WMO-specified agency in charge of marine forecasts for the area. They in turn, contacted the “Centro Integrado de Meteorologia e Recursos Hidricos” for the State of Santa Catarina, who contacted the NHC directly for additional information. These channels remained open through the remainder of the life of the cyclone.

337

Most procedures used by the NHC for North Atlantic tropical cyclones were activated for Catarina. The TAFB, SAB, and AFWA provided satellite intensity estimates, while the Naval Research Laboratory in Monterey, California activated its web site for the collection and distribution of microwave satellite imagery. The CIMSS activated its web site for the collection and distribution of satellite-derived winds and the products derived from them. The NHC ran its suite of hurricane track model guidance on Catarina, including the Geophysical Fluid Dynamics Laboratory (GFDL) model, a high-resolution regional model specifically designed for tropical cyclone forecasting. This was made possible by programmers at the NHC, GFDL in Princeton, New Jersey, and the National Centers for Environmental Prediction in Washington, DC. The results of the model runs were communicated to the meteorologists in Brazil Global models provided reasonable forecasts of the large-scale characteristics of the Catarina event. However, the global model forecasts underestimated the intensity of the system and the forecasted trajectory varied significantly depending on the model and on the initial perturbation in the ensemble forecasting (Silva Dias et al. 2004). Figure 2.a.18 shows the a comparison of the 48 hour precipitation forecast valid at 28 March 2004 at 00 UTC produced by the CPTEC global model (http://www.cptec.inpe.br/prevnum/ exp_global.shtml) at the resolution of the global ensemble forecast (T126), an experimental ensemble resolution (T170), the operational high resolution global forecast (T254) and an experimental high resolution global forecast (T515). These results are discussed by Bonatti et al. (2006). It is clear that increased resolution plays a significant role in the positioning of the rain area and its intensity. The very high resolution (equivalent to approximately 27km at the latitude of the storm ) was actually able to forecast the total accumulation quite reasonably. However, pressure gradients and wind were severely underestimated (not more than about 15m/s) at landfall. Regional models run by the weather services with horizontal resolutions between 20 and 40 km were also unable to accurately forecast the development of the vortex and its westward movement. The central pressure in most of the forecasts remained above 1000 hPa. The operational CPTEC ETA model at the time with horizontal resolution of 40km (Figure 2.a.19) also under-estimated the intensity of the low pressure during the life cycle of the storm. Only under 90h forecast lead time were the forecasts able to properly capture the trajectory of the system. NCEP operational forecasts with the global model were also available at the time for operational use with higher resolution. These forecasts also underestimated the intensity of the system and there was significant discrepancy concerning the trajectory and the final stage, before landfall (Silva Dias et al. 2004). The GFDL model (Bender and Marchock, 2004) forecasted many important features of the hurricane, including its path, with great accuracy (not shown) due to the initialization approach. Conceptual models such as the one proposed by Conway (1997), qualitatively adapted for the Southern Hemisphere, can be very useful in conjunction with satellite and other remote sensing data such as weather radars. The initial forecasts were based on model outputs and climatology with less emphasis to satellite data and conceptual models. Hurricane Catarina indicated an urgent need for more specific training programs, observing platforms and procedures to analyze and to forecast unusual and highly destructive weather systems 2.a.9 Catarina and Global Warming

At this stage there is no agreement as to whether Catarina is or is not related to climate change. Pezza and Simmonds (2005) suggested the former possibility in linking Catarina with a very unusual large scale pattern which had a character very similar to the highly significant observed increase in the positive phase of the most important mode of circulation in the Southern Hemisphere, the Southern Annular Mode (SAM) (Cai et al, 2005). However, more research is clearly needed. However, the problem seems to be significantly more complex than a simply direct change in the wind shear pattern over the region. Recent research has also shown that there has been a very significant increase in the positive phase of the SAM (Marshall, 2003) which has been attributed partially to ozone

338

losses (Thompson and Solomon, 2002, Cai, 2006) and to an increase in greenhouse gases (Fyfe et al 1999). If global warming is changing the SAM (Cai et al 2006) it could then indirectly change the hurricane formation in the South Atlantic Ocean via the possible links described above, even without directly making the SSTs over the South Atlantic significantly warmer, as seen during March 2004. Although this may be still speculative, it is also physically consistent with the slight negative trend (though not statistically significant) in the wind shear for the area of interest to the south of where Catarina was formed. 2.a.10 Recommendations Hurricane Catarina represents a major mark in South America meteorology since it prompted the weather services to keep an open perspective on the different types of weather systems Earth’s system can produce on a time of fast climatic changes. Important scientific and operational questions arose after March 2004. First, what Catarina really was and how should we refer to it? Second, was Catarina a result of natural climate variability only, or could it also be related to climate change due to anthropogenic influences? There is speculation on the relationship between Catarina and global climate change. Since the record of tropical and subtropical cyclones in the South Atlantic is somewhere between incomplete and non-existent, this will likely remain speculation for now. However, Haarsma et al (1993) showed results of global climate modeling suggesting that South Atlantic tropical cyclones could occur – or increase in frequency – in a world with doubled atmospheric CO2. Finally, from an operational point of view, what should be done to improve the predictability of this type of weather system?

There remains a question of what should be done if storms like Catarina occur again. It appears unlikely that such storms will become a frequent enough occurrence to warrant developing the elaborate WMO-organized program present in other basins. Holland (2005) recommended that the National Institute of Meteorology (INMET) of Brazil train a group of forecasters in the methods of tropical cyclone analysis and forecasting, so when an event like Catarina occurs again these forecasters will be able to put aside their normal duties and forecast the cyclone. While the final form of a potential tropical cyclone warning service for the South Atlantic remains unresolved, efforts are underway between the TPC and INMET for such training to occur. The special workshop on the Catarina event held in June 2005 at the Center for Weather Forecasting and Climate Research in Brazil, organized by the Brazilian Meteorological Society to the following recommendations: • Organize an Alert Center for Severe Storms, nationally distributed with experts that can be called to advice the operational forecasting centers in emergency situations; • Investment in capacity building for transferring meteorological information to the media; • Incentives for the exchange of specialized personnel between national and international institutions with expertise on the analysis and prediction of intense cyclones; • Improve the observational system in the ocean (increasing the number of observational buys) and over land (improving the reliability of the regular operation and dissemination of the information provided by the weather radar system); • Improve the data dissemination system and training in the use of Remote Sensing Techniques applied to the monitoring of intense cyclones; • Improve the frequency of geoestacionary satellite data information in the Southern Hemisphere either through the development of national programs or in association with the NOAA and the European Satellite Agency;

339

• Articulated solution with WMO should be sought in order to improve the personnel training and dissemination of the information. In addition, the workshop recommended that in spite of the hybrid structure of the cyclone, if another case arises in the future, it is recommended that the systems be called “hurricane” when the estimated wind speed falls in the category I. This recommendation is based on the fact that the public perception that this type of event is significantly different from the regular extratropical and subtropical lows that regularly reach the southern part of Brazil . 2.a.11 References

Bender, M. A. and T. Marchok, 2004: A summary of upgrades to the operational GFDL hurricane model for 2003. 26th Conf. on Hurricane and Tropical Meteorology, Miami. Paper 10C.3.

Bister, M., and K. A. Emanuel, 2002: The genesis of Hurricane Guillermo: TEXMEX analyses and a modeling study. Mon. Wea. Rev., 125, 2662-2682.

Bonatti J.P., Rao, V.B., 1987: Moist baroclinic instability of North Pacific and South American intermediate-scale disturbances. Journal of the Atmospheric Sciences, v. 44, p. 2657-2667.

Bonatti, J. P. ; Rao, V. B. ; Silva Dias, P. L.,2006: On the westward propagation of the Catarina Storm. In: 8th International Conference on Southern Hemisphere Meteorology and Oceanogrphy, 2006, Foz do Iguaçu, PR. Proceedings of 8 ICSHMO. p. 1659-1675.

Brueske, K. F., and C. Velden, 2003: Satellite-Based Tropical Cyclone Intensity Estimation Using the NOAA-KLM Series Advanced Microwave Sounding Unit (AMSU). Mon. Wea. Rev., 131, 687-697.

Cai, W., Shi, G., Cowan, T., Bi, D., Ribbe, J., 2005: The response of the Southern Annular Mode, the East Australian Current, and the Southern mid-latitude ocean circulation to global warming. Geophysical Research Letters, 32, doi:10.1029/2005GL024701.

Cai, W., 2006: Antarctic ozone depletion causes an intensification of the Southern Ocean super-gyre circulation. Geophysical Research Letters, 33 doi:10.1029/2005GL024911

Climanálise, 2004: Boletim the Monitoramento e Análise Climática. CPTEC, February 2004 (available at http://www.cptec.inpe.br/products/climanalise/0204/index.html )

Conway, E. D. and MSGC, 1997: An introduction to satellite image interpretation. The Johns Hopkins University Press. 242 pp.

Davis, C.A., and L. F. Bosart, 2003: Baroclinically Induced Tropical Cyclogenesis, Mon. Wea. Rev., 131, 2730–2747.

Davis, C., and L. F. Bosart, 2004: The TT problem. Bull. Amer. Meteor. Soc., 85, 1657-1662.

Dvorak, V. E., 1984: Tropical cyclone intensity analysis using satellite data. NOAA Tech. Rep. NESDIS 11, National Oceanic and Atmospheric Administration, Washington, DC, 47 pp.

Emanuel, K. A., 1999: Thermodynamic control of hurricane intensity. Nature, 401, 665-669.

Franklin, J. L., 2006: Tropical Cyclone Report – Hurricane Epsilon. Available from the web site of the Tropical Prediction Center at URL www.nhc.noaa.gov/pdf/TCR-AL302005_Epsilon.pdf

Fyfe, J. C., G. J. Boer, and G. M. Flato, 1999: The Arctic and Antarctic Oscillations and their projected changes under global warming, Geophys. Res. Lett., 26, 1601-1604.

Gan, M. A. and V. B. Rao, 1991. Surface cyclogenesis over South America. Mon. Wea. Rev.,19 (5):1293-1302.

340

Gevaerd, R., M. Longo, P.L.Silva Dias e F.V. Branco, 2004: Análise da Precipitação Associada ao Ciclone Catarina. In: XIII Congresso Brasileiro de Meteorologia,2004, 0780: Fortaleza, 10pp.

Haarsma, R. J., J. F. B. Mitchell, and C. A. Senior, 1993: Tropical disturbances in a GCM. Climate Dyn., 8, 247-257. Hart, R.E., 2003: A cyclone phase space derived from thermal wind and thermal asymmetry. Mon. Wea. Rev., 131, 585-616

Hebert, P. J., and K. O. Poteat, 1975: A satellite classification technique for subtropical cyclones. NOAA Technical Memorandum NWS SR-83, U.S. Dept. of Commerce, National Weather Service, Ft. Worth TX, 25 pp.

Holland, G., 2005: Comparison between Catarina and other tropical-subtropical systems. Workshop on the Catarina Phenomena. Available at: http://www.sbmet.org.br/publicacoes/informativo/2005_07 /index.html .

Joyce, R. J., J. E. Janowiak, P. A. Arkin, and P. Xie, 2004: CMORPH: A method that produces global precipitation estimates from passive microwave and infrared data at high spatial and temporal resolution. J. Hidrometeor., v.5. p. 487-503.

Machado, L., A., T.; Laurent, H. ,2004: The convective system area expansion over Amazonia and its relationships with convective system life duration and high-level wind divergence. Monthly weather review, v. 132, n. 4, p. 714-725.

McAdie, C. J. and E. N. Rappaport, editors, 1991: Diagnostic report of the National Hurricane Center, June and July 1991. Volume 4, Number 1, 10-14. Available from the Tropical Prediction Center/National Hurricane Center. McTaggart-Cowan, R., L. F. Bosart, C. A. Davis, E. H. Atallah, J. R. Gyakum, and K. A. Emanuel, 2006d: Analysis of Hurricane Caterina (2004). Mon. Wea. Rev., in press.

Marshall, G. J., 2003: Trends in the Southern Annular Mode from Observations and Reanalyses, J. Clim., 16, 4134-4143.

Nakano, M. and K. Nakajima, 2004: Influence of vertical wind shear on the initial vortex formation of tropical cyclone. 26th Conf. on Hurricane and Tropical Meteorology, 2004, Miami, AMS, Paper P1.49.

Pereira Filho, A. J. ; Lima, Raquel Silva . Synoptic and mesoscale analysis of hurricane Catarina. In: 8th International Conf. on Southern Hemisphere Meteorology and Hydrology, 2006, Foz do Iguaçu. 8th International Conf. on Southern Hemisphere Meteorology and Hydrology. São José dos Campos : CPTEC/INPE, 2006.

Pezza, A.B., and Simmonds, I, 2005: The first South Atlantic hurricane: Unprecedented blocking, low shear and climate change. Geophysical Res. Letters 32, doi:10.1029/2005GL023390.

Pezza, A.B. and I. Simmonds, 2006. Catarina: The first South Atlantic hurricane and its association with vertical wind shear and high latitude blocking. In: International Conference on Southern Hemisphere Meteorology and Oceanography (ICSHMO), 8., 2006, Foz do Iguaçu. Proceedings... São José dos Campos: INPE, 2006, p. 353-364. CD-ROM. ISBN 85-17-00023-4.

Reale, O. and R. Atlas,2001: Tropical cyclone-like vórtices in the extratropics: observational evidence and synoptic analysis. Weather.Forecasting,Vol. 16, 7-34.

Silva Dias, P.L., M.A. F. Silva Dias, M. Seluchi e F.A. Diniz, 2004: O Cilclone Catarina: Análise Preliminar da Estrutura, Dinâmica e Previsibilidade. In: XIII Congresso Brasileiro de Meteorologia,2004, 0764: Fortaleza, 10pp.

Silva Dias, P.L. and M. Longo, 2005: Catarina Cyclone Forecast Uncertainties: role of latent heat release. In http://www.sbmet.org.br/publicacoes/informativo/2005_07/ index.htm

341

Menezes, W.F., P.L.Silva Dias, 2004: Um Estudo do Impacto das Opções Físicas do Modelo RAMS na Simulação Numérica do Ciclone Catarina. In: XIII Congresso Brasileiro de Meteorologia,2004, 0586: Fortaleza, 10pp.

Silva, R.R., P.L. Silva Dias, A.W. Gandu, D.S. Moreira, 2004: Impactos da Temperatura da Superfície do Mar no Ciclone Catarina. In: XIII Congresso Brasileiro de Meteorologia,2004, 0447: Fortaleza, 10pp.

Thompson, D. W., and S. Solomon, 2002: Interpretation of Recent Southern Hemisphere Climate Change, Science, 296, 895-899.

342

Figure 2.a.1 300 hPa geopotential height from March 20 to 31, 2004 at 12h intervals.

343

Figure 2.a.2 GOES Infrared imagery for the Catarina cyclone in March 2004: a) 03 UTC 22 March, b) 15 UTC 22 March, c) 03 UTC 23 March, d) 15 UTC 23 March, e) 03 UTC 24 March, f) 15 UTC 24 March, g) 03 UTC 25 March, h) 15 UTC 25 March, i) 00 UTC 26 March, j) 15 UTC 26 March, k) 15 UTC 27 March and l) 15 UTC 28 March.

344

Figure 2.a.3 Catarina Hurricane trajectory in perspective with the surrounding maximum Sea Surface Temperatures (SSTs). The South American sector is showing: I) 2 km resolution topography plotted for elevations above 500 m, with darker yellow tones indicating elevations above 1500 m; II) Tropical cyclone Catarina’s trajectory as derived from the University of Melbourne automatic tracking algorithm showing the central locations every 06 hours and III) Maximum SSTs (oC) for the period between the 20th and the 28th of March. The date and hour (UTC) are indicated next to the trajectory for some selected periods. From Pezza and Simmonds (2005)

345

Figure 2.a.4 Vertical structure of the Catarina Cyclone based on: (a) AVN (1o resolution analysis) and (b) AMSU soundings ((http://moe.met.fsu.edu/cyclonephase/).

346

Figure 2.a.5 Vertical cross section of vorticity (shaded) and vertical motion (contour) and relative humidity (shaded) and potential temperature (contour) at the latitude of the strom center on (a) 26 March 12 UTC and (b) 27 March 2004, 12 UTC. The white line indicates the longitude of the storm center. (from Bonatti et al. 2006)

347

Figure 2.a.6 AMSU-derived azimuthally-averaged temperature anomalies for the pre-Catarina low

near 2100 UTC 23 March 2004. Image courtesy of the Cooperative Institute for Research in the Atmosphere (CIRA) at Colorado State University.

348

Figure 2.a.7 Upper-level winds derived from GOES-12 imagery near Catarina at 1800 UTC 27 March. Image courtesy of CIMSS.

349

Figure 2.a.8 High resolution GOES infrared imagery at land fall. Convective development is seen right at the center of the system. Another intense convective cell is observed in the southeastward sector of the eye wall. (contribution of Dr. Conrado Conforte)

350

Figure 2.a.9 Radar cross section from the TRMM satellite at 11 UTC 27 March 2004. Image courtesy of NASA.

351

Figure 2.a.10 Time evolution of hourly measurements of pressure (right hand scale) and wind (in red, left hand scale) between 26 and 29 March and station level pressure (in blue, right hand scale) from the 26th of March at 0:00 local time to March 29th at 00:00 local time for São Bento (28o 36’S, 49o 33’W, 135 m). (From Information Center of Environmental Resources and Hydrometeorology of the State of Santa Catarina in Brazil –http://ciram.epagri.rct-sc.br). (from Pezza and Simmonds 2006)

352

Figure 2.a.11 Histogram of western South Atlantic dipole blocking duration in the NCEP/NCAR Reanalysis from 1971-2001. Dipole blocking is defined as anomalously low 500 hPa heights in the equatorward box of the inset combined with anomalously high heights in at least two of the three poleward boxes. (Adapted from McTaggart-Cowan 2006.)

353

Figure 2.a.12 Environmental Vertical Wind Shear (200/850 hPa) anomaly (m/s) averaged over the 23rd – 28th March 2004 period. The whole Catarina’s trajectory as seen in Figure 3 is also indicated. The –15 m/s wind shear anomaly near the southern Brazilian coast approximately corresponds to an average wind shear value of 8 m/s, which is the ideal threshold for EVWS in hurricanes. From Pezza and Simmonds (2005).

354

Figure 2.a.13 Time series of the blocking index B at 700 hPa and the Environmental Vertical Wind Shear (EVWS) index for all March months during 1979-2004. The horizontal red line indicates the maximum B index reached five days prior to Catarina’s genesis (upper arrow), and the blue line indicates the minimum wind shear reached during the Tropical Transition phase (lower arrow). A 1-2-1 time filter was applied twenty times in order to eliminate the very high frequency variations. Data plotted every six hours. From Pezza and Simmonds (2005)

355

Figure 2.a.14 Outgoing longwave anomalies (OLR) in January (a), February (b) and March (c) of 2004. (data from CPTEC homepage: www.cptec.inpe.br ).

356

Figure 2.a.15 Time series and real part of the wavelet spectrum of: (a) amplitude of a measure of the Madden Julian Oscillation at 40oW, (b) Southern Oscillation Index, (c) Antarctic Annular Oscillation. (Contribution of Marcelo Schneider).

357

Figure 2.a.16 Precipitation rate (mm/3h) obtained with the downscaling of the NCEP operational 1o analysis to a 8 km grid with BRAMS. Wind speed in ms-1. (Silva Dias et .al 2005).

358

Figure 2.a.17 Sensitivity of the mesoscale downscaling (8km) of the NCEP analysis to changes in the vertical profile of the heating (intensification and change of the vertical gradient) and domain size (larger and smaller domains).

359

Figure 2.a.18 Accumulated precipitation at 27 March, 2004 at 00 UTC based on the CPTEC global model forecast at T126 (resolution of the operational ensemble forecasting), T170 (experimental ensemble forecasting); T253 (operational global model forecast) and T515 (experimental high resolution global forecast). (from Bonatti et al. 2006)

360

Figure 2.a.19 Regional operational forecasts produced by the Center for Weather Forecasting and Climate Research/CPTEC with the ETA model at 40km horizontal resolution. Pressure at sea level is shown in contour lines (hPa), wind in vectors and precipitation in shading. (from Silva Dias et al. 2005)

361

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES

Session 2 b : Special Focus Session THORPEX: a focus on tropical cyclone related research Rapporteur: J. D. Abraham Environment Canada 45 Alderney Drive, Dartmouth, NS Canada Email: jim.abraham@ec.gc.ca fax: 902-426-9158 Working Group: D. Parsons, P. Harr, N. Kitabatake, S Majumdar Note: This report is based largely on the Scientific Program Overview and Experimental Design documents for the THORPEX Pacific-Asian Regional Campaign prepared by David Parsons and colleagues: http://www.ucar.edu/na-thorpex/ 2b 1.0 Introduction As recognition of this increasing value to society of improvements in weather prediction skill, the ~180 nations of the 14

th World Meteorological Congress established the THORPEX (The Observing System

Research and Predictability Experiment) research and development program. Specifically, THORPEX is a component program of World Weather Research Program within the World Meteorological Organization (WMO/WWRP). The THORPEX Science (Shapiro and Thorpe 2004) and Implementation Plans (EG-TIP 2005) outline an ambitious research agenda designed to accelerate improvements in the accuracy of 1-day to two-week forecasts of high-impact weather for the benefit of society, the economy, and the environment. THORPEX also seeks to improve the utilization of weather products, since, for example, the mitigation and response to a weather disaster also depend on decision-making, communication, allocation of existing resources, and the ability of governments and aid organizations to mobilize a response. The international component of THORPEX results from the realization that forecast improvements on the 1 to 14-day time-scale rest on global prediction models and the global observing system. The core objectives of the international THORPEX program (see Shapiro and Thorpe 2004 and http://www.wmo.int/thorpex) include advancing knowledge of the global-to-regional influences on the initiation, evolution and predictability of high-impact weather and contributing to the development of advanced data assimilation and ensemble prediction systems, and to the design of the future global observing system. The core objectives include international field campaigns focused on regional forecast problems facing Africa, Asia, Europe, North America and the Southern Hemisphere. For example, North America experiences, with some frequency, a variety of weather disasters with individual events that cause damage over a billion dollars and present a significant risk to public safety. Such weather events include floods, the dry gusty conditions of fire weather, severe convection outbreaks, tropical cyclones (TCs), and winter storms capable of freezing rain, high winds, snow, freezing rain, and heavy rainfall. For the Asian nations of the Pacific Rim, a key forecast problem is to better mitigate and respond to heavy rainfall and TCs, since these events are their most damaging natural disasters. Recent research clearly points to the western North Pacific playing an important and unique role in defining many characteristics of the middle latitude circulation of the Northern Hemisphere. Over the western and central North Pacific, baroclinic energy conversion generates a large amount of kinetic energy that is instrumental in maintaining the storm tracks downstream over the eastern North Pacific, North America, and North Atlantic (Chang and Yu 1999; Orlanski and Sheldon 1995; Nielsen-Gammon and Lefevre 1996; Danielson et al. 2004). This implies that many of the high-impact weather events

362

that occur over North America have a dynamical origin upstream over the western North Pacific basin. Furthermore, forecasts of downstream developments in the storm track over the eastern North Pacific that impact western North America often contain large errors (McMurdie and Mass 2004). Therefore, it is hypothesized that improved treatment and increased understanding of the dynamical linkages between the development of high-impact weather events that occur over North America to specific weather systems upstream over the western North Pacific will lead to a significant increase in forecast skill of these downstream events. Indeed the calculation of average forecast sensitive areas for prediction of high-impact weather events over North America for the time-scales ranging from ~2 to 5 days show a strong sensitivity to conditions over the western and central portions of the North Pacific basin (e.g., see Figure 4a, b of Reynolds and Gelaro 2001). Forecasts of weather events over Alaskan and portions of the Canadian Arctic also show a strong sensitivity to conditions over the North Pacific (e.g., Szunyogh 2002). One mechanism by which events over the western North Pacific may trigger downstream responses over the eastern North Pacific and North America is via upper-tropospheric wave packets (Fig. 2.b.1). Hakim (2003) and Chang (2005) have provided evidence that wave packets on the primary Asian waveguides increase the likelihood of the development of intense cyclones over the North Pacific. There are also indications that the wave packets are, in turn, invigorated by the cyclogenesis events, which makes their impacts farther downstream over North America potentially more significant. Hakim (2003 and 2005) demonstrated that upper-tropospheric, eastward-propagating wave packets are a dominant source of forecast errors over the North Pacific. The forecast error patterns move with an eastward group velocity of about 30

o-40

o per day, which means that the leading edge of increased

forecast error can reach western North America in about 3 days and the Great Lakes region in 4-5 days. An example of Rossby wave trains apparently generated by high-impact events over the western Pacific associated with the lifecycle of TCs is shown in Fig. 2.b.1. These wave trains were associated with three major weather disasters on the western coast of the US. Thus, accurate short-range (<3 days) predictions of aspects of the lifecycle of a TC near the east Asian coast will mean an increased likelihood of accurate medium range (3-7 day) predictions of floods, wide-spread severe weather outbreaks and damaging extratropical cyclones downstream over North America. This connection is the underlying focus of the THORPEX Pacific-Asian Regional Campaign (PARC).

Fig. 2.b.1 a) Time-longitude diagram of 250 hPa meridional winds (m s

-1) from 0000 UTC 5 October

2003 – 1200 UTC 31 October 2003 (Figure made at http://www.cdc.noaa.gov). The diagonal dashed lines highlight eastward-moving, upper-tropospheric wave packets that originated over eastern Asia. Three relatively poorly predicted major west coast weather events are shown. b) Integrated water vapor on the 17

th of October coinciding with the record flood in British Columbia showing the tropical moisture

plume from the western Pacific being advected into the region with this first Rossby wave train. c)

363

Outgoing longwave radiation for the 21 October 2003 showing four tropical convective events (at this time one typhoon, one tropical storm and two tropical depressions) in the region of the genesis/intensification of the Rossby wave trains. The red arrow shows the propagation wave train from the tropical convection as suggested in the OLR for the western Washington flood. All figures are constructed from the NOAA/CDC website and provided by Dave Parsons, NCAR based on analysis by Melvyn Shapiro (NOAA), Lynn McMurdie (U Washington) and Parsons. The discussion that follows, focuses on the scientific challenges and objectives, as well as the experimental design within the proposed THORPEX: Pacific-Asian Regional Campaign. Many of these challenges are common to other basins. However, for the purposes of this special session, PARC serves as the best example of a THORPEX research initiative focused on tropical cyclones. 2b 2.0 Scientific Challenges associated with Predictability of High-Impact Weather There are a number of scientific challenges associated with advancing knowledge of high-impact weather in the vicinity of the Asian Pacific rim, including the generation of downstream high-impact events over North America and the Arctic. These challenges include advancing knowledge and the impact on forecast skill of: 1. genesis and evolution of tropical cyclones 2. extratropical transitions 3. upper tropospheric wave trains Improved forecasting accuracy for all these events would have significant societal benefits. 2b 2.1 Advancing knowledge and the predictability of the genesis and evolution of tropical cyclones One of the primary foci of the THORPEX Asian Regional Committee is a strong need for accurate predictions of TC track and landfall with the goal of improved prediction and extending the forecast lead- time. The life cycle of a TC may be divided into several phases, including cyclogenesis, development and motion of a mature cyclone, the subsequent extratropical transition and/or landfall. Each phase brings a new spectrum of issues related to physical and dynamical processes, and their predictability. Only a small fraction of tropical disturbances eventually become TCs. The challenges of advancing understanding of the dominant mechanisms and factors that limit the predictability of tropical cyclogenesis in a broader context have recently received increasing attention in both research and operational prediction. For example, the leading scientists from the research community and operational centers that comprise the WMO/CAS Working Group on Numerical Experimentation (WGNE) recommended that the operational centers provide statistics on genesis. Accurate forecasts of genesis would open up the possibility of greatly increasing the time scales of when typhoon predictions have utility with a variety of benefits (e.g., ocean shipping, energy production, planning and carrying out evacuations, rescue and early recovery efforts especially for urban centers, organization of international aid) as well as improving downstream predictions. Initial studies have shown some level of skill at relatively long lead times, but this research also raise questions about false alarms and missed events. The ability of numerical models to accurately represent physical quantities relevant to genesis such as warm core magnitude, the vertical wind shear, convective instability and mid-level moisture requires assessment. Unfortunately, the lack of detailed data sets and hence quality of analysis compromises our ability to evaluate critical aspects of model performance and confirm or deny current hypotheses. The combination of high-resolution, cloud-resolving models in conjunction with in-situ observational data collected during PARC is necessary to address the issues on the processes, scale interactions and predictability of tropical cyclogenesis.

364

2b 2.2 Advancing knowledge of extratropical transitions and their impact on forecast skill One aspect of the Asian THORPEX interest in tropical cyclones is the improved understanding and prediction of recurvature and extratropical transition (ET), which are critical forecast problems for Japan, Korea and parts of China. This objective is a key link between the North American and Asian THORPEX studies in PARC and also with the International Polar Year (IPY), because ET storms also directly impact the Arctic as they move northward and such storms have dramatic meteorological effects downstream. The poleward movement of a TC in the western North Pacific during summer and fall leads to an interaction with the midlatitude baroclinic zone and jet stream that can result in an extratropical transition. The primary scientific issues associated with ET and downstream impacts due to ET events may be placed in a framework of mechanisms, predictability, and strategies for increasing predictability. Periods of low predictability are expected from uncertainty theory because the ET involves complex physical interaction between the tropical cyclone and the mid-latitudes. Furthermore, during the transformation stage from a TC to the initial development of an extratropical cyclone, phenomena with high-societal impacts include strong winds, high waves, and heavy precipitation. ET events can trigger upper-level wave trains that propagate faster than the ET event. The dynamic and physical properties during the ET that lead to this wave train are not well known. Furthermore, propagation of the wave train downstream from the ET event may trigger high-impact events over North America. 2b 2.3 Advancing knowledge of upper-tropospheric wave trains and their impact on forecast skill Recent research, based on modern statistical techniques applied to reanalysis data sets, diagnostic techniques applied to numerical weather forecasts, satellite imagery, idealized model experiments, and case studies with state-of-the-art numerical weather prediction models, suggests that the western North Pacific region plays a unique role in shaping the circulation of the Northern Hemisphere (NH) extratropics. The evidence for this relationship includes: i) the role of the maintenance of the storm track by the generation of a vast amount of kinetic energy by baroclinic energy conversion in the western and central North Pacific ii) studies of mid-latitude analysis and forecast errors suggest that errors in the western North Pacific spread downstream at a speed of about 30 m s

-1, with their effects reaching the U.S. and

Canada within a few days iii) studies into the skill of North American and Arctic forecasts show a strong sensitivity to areas over the NW Pacific. Many forecast failures in the prediction of Northern Hemisphere extratropical storms have their roots in the western and central North Pacific. The number of forecast failures is significant, as shown by McMurdie and Mass (2004), with average errors, for example, in the 48-h prediction of west coast winter cyclones measured in terms of 100s of kilometers. The proposed research aims to reduce the number of these forecast failures by improving the analysis and modeling of the processes that initiate and maintain the primary cyclogenesis in the western and central North Pacific and of those that propagate these effects downstream 2b 3.0 Scientific Objectives of PARC PARC is THORPEX’s first scale-interaction experiment and perhaps the community’s first field effort focused to a large degree on medium range weather phenomena and prediction. The experiment brings together research facilities from Asia, Europe and North America with participation ranging from

365

the academic research community to scientists at the operational centers. We envision the PARC experiment to be embedded within a longer-term observational, theoretical and numerical THORPEX research focus on the scientific goals associated with PARC. An overall important scientific goal includes the improved understanding of the dynamics and factors that limit the predictability of downstream high-impact weather events as discussed in 2b Section 2.0 above. Some other specific objectives include: i) Developing, improving, and evaluating data assimilation strategies; understanding error growth and scale interactions; and the improved utilization of satellite measurements, ii) Testing the improvement in local and downstream forecast skill afforded by high-resolution, non-hydrostatic modeling; iii) Testing new strategies and observational systems for adaptive observing and modeling strategies; iv) Improving the interpretation and utility of ensemble forecast systems; and v) Understanding and improving society’s response to weather disasters, including the appropriate use and evaluation of probabilistic information. 2b 3.1 Developing, improving, and evaluating data assimilation strategies; understanding error growth and scale interactions; and the improved utilization of satellite measurements Improved assimilation of satellite observation data is essential to PARC (and overall THORPEX) goals for deterministic and ensemble weather prediction, and also for establishment of high-quality atmospheric analyses required for climate monitoring. The Pacific basin is a region of particular importance for satellite data assimilation because it is a vast area in which in-situ observations are sparse or non-existent, and it produces a wide variety of high-impact weather events that affect adjacent heavily populated areas of Asia and North America. Recent and ongoing increases in the amount, quality, and variety of satellite atmospheric observations represent a tremendous opportunity for progress in numerical weather prediction, including improved local and downstream forecasts of devastating TCs, subsequent ET events and intense winter cyclogenesis. Currently, space-based observations represent more than 80 percent of all data assimilated for operational numerical weather prediction. The ultimate goal is assimilation of satellite data at the scale of the observations. Progress towards this goal depends on improved techniques for satellite data assimilation and on increases in computational capability. Whereas developing new satellite technology represents an investment of billions of dollars by many nations, only a very small percentage of this large technology investment has been spent on research to make better use of the satellite data that are provided. THORPEX can accelerate progress in satellite data assimilation that is needed to improve weather prediction and climate monitoring. Recently developed four-dimensional assimilation methods, such as the Ensemble Kalman Filter and 4D-Var, have new capabilities for the description of developing dynamical features. An understanding of the inherent qualities and limitations of each method may be gained via a combination of Observation System Simulation Experiments (OSSEs), Observing System Experiments (OSEs), and operational frameworks. PARC will provide a focal point for the community to conduct experiments to advance data assimilation techniques through testing these advances on a challenging set of problems with significant societal and economic benefits for a large portion of the world’s population. The DA issues that require attention include: (a) Cycling of flow-dependent error covariance matrices. (b) Model uncertainty.. (c) Multi-scale observations and increments. (d) Correlated observation error statistics.. (e) Linearity. Conventional DA methods assume Gaussian error statistics and linear dynamics

366

2b 3.2 Testing the improvement in local and downstream forecast skill afforded by global high-resolution, non-hydrostatic modeling PARC is associated with the challenging phenomena of TC and extratropical transition events and how these events and other systems excite of these Rossby wave trains associated with downstream high-impact weather. Another source of Rossby wave excitation is the deep organized convection associated with Madden Julian Oscillation (MJO)/ El Niño–Southern Oscillation (ENSO) events (see Ferranti et al. 1990). The ENSO/MJO effect the frequency of TCs in both hemispheres (e.g., Keen 1982; Lander 1990) with some recent suggestions that TCs can also impact the ENSO events (Sobel and Camargo, 2005). The MJO is generally poorly simulated in weather and climate models. To capture the MJO, the modulation of deep convection and the global dispersion of Rossby waves, as well as the downstream high-impact events, the ideal situation would be to have a global model with cloud-resolving capabilities. Simulations and sensitivity experiments are planned using high-resolution numerical models, theoretical studies, and comparison of model results with observations and analyses. In particular, the Canadian GEM model will be used to simulate cases of MJO and radiation of Rossby wave packets. In view of these high-resolution simulation studies, an unprecedented convective-scale to continental-scale simulation will be done on the Earth Simulator Center (ESC). This simulation and others with lower resolution will be used to examine the expected implication of MJO for numerical weather forecasting. More specifically, diagnostics studies, such as Charron and Brunet 1999) and sensitivity experiments will be made to determine the important parameters that affect MJO and Rossby wave radiation that affect the North American and European regions. 2b 3.3 Testing new strategies and observational systems for adaptive observing and modeling strategies It is important to build on previous targeting results, including such operational programs in the US for winter storms and hurricanes, and the previous Atlantic THORPEX regional program, when deploying resources to investigate adaptive sampling strategies during PARC. Three general topics arise concerning adaptive targeting during PARC. (1) Adaptive targeting of tropical cyclones for improved prediction: A focus of the THORPEX Asian Regional Committee for PARC on adaptive measurements for typhoon prediction is prompted by successful dropwindsonde observations in the synoptic environment of typhoons (Wu et al. 2004; 2005) and hurricanes (Aberson 2003) to improve operational forecasts. Studies such as Aberson (2003) have shown that incorporation of dropwindsonde data around TCs approaching landfall has significantly increased the rate of increase in forecast skill. Although these aircraft-borne observations have been shown statistically to improve the accuracy of global model track forecasts (i.e., the average improvement typically ranges between 20 and 30%), the scientific premise of how observations influence forecasts of TC motion and structure remains unexplored. Another basic research topic to be explored is increase the scientific understanding of the significant differences in the prediction of targeting locations by different targeting techniques (Majumdar et al. 2006). The techniques include sensitivity to conditions near the cyclone, but sometimes also include features within the middle latitudes. For the first time in a targeting experiment, we have the opportunity to oversample to better understand the impact of following different targeting experiments. Moreover, the benefits of assimilating additional observations from different platforms, including airborne and satellite remote sensing, on multiple spatial and temporal scales using novel data assimilation methods have not been studied. For example, studies will be provided to determine the relative contributions from in-situ adaptive measurements versus new satellite systems such as COSMIC or whether complementary approaches (in-situ winds only) should be developed. (2) Impact of adaptive measurements for tropical cyclones on the predictability of downstream events: ET events are a major source of downstream uncertainty as well a trigger for the downstream

367

generation of high-impact weather. Since research to date suggests that track forecasts of tropical system will improve through adaptive measurements, the question arises as to whether adaptive measurements taken to improve typhoon track prediction will improve the prediction of ET and downstream effects? Such an improvement is expected, since one would expect improvement in the prediction of where and when a TC enters the middle latitude westerlies. For this reason adaptive measurements will be taken during the recurvature of the TC. Alternate hypotheses are that the uncertainty lies with the middle latitude westerlies, a lack of resolution of the vortex moving into the westerlies or in the cyclogenesis process itself. (3) Adaptive techniques for middle latitude events: The case for in-situ targeting in winter storms is the subject of active debate, since the results are generally positive, but with the exception of a few events, relatively weak. Many of the issues associated with targeting of middle latitude systems are discussed in Langland (2006), Shapiro and Thorpe (2004) and in the report from the THORPEX Data Assimilation and Observing System Working Group (Rabier 2006). 2b 3.4 Improving the interpretation and utility of ensemble forecast systems Since the early 1990s, ensembles of global model forecasts have been issued daily at individual operational centers. Assuming that errors in the initial conditions and numerical models are well captured, the ensemble provides an estimate of uncertainty of a particular forecast. For example, a tight cluster of TC track forecasts provided by the ensemble increases forecaster confidence in the prediction, thereby reducing the necessary area of coastal evacuation, mitigating costs and increasing public confidence. In contrast, a wide spread of TC track forecasts necessitates the evacuation of a large stretch of coastline, even though the majority of the evacuation area will likely be relatively unscathed. In addition to the prediction of uncertainty, the weighted average of a multi-model ensemble of forecasts has generally proven to be more accurate than a single forecast, even of higher resolution. Currently, only a limited number of forecasts (10-100) have been considered in the ensemble, and partially because of this, more sophisticated applications than those given above have not been tested. The implementation of a multi-center, bias-corrected ensemble hosted in North America, known as the North American Ensemble Forecast system (NAEFS), to be integrated within the THORPEX Interactive Grand Global Ensemble (TIGGE), will open new possibilities for advanced ensemble-based research and forecasting applications based on several thousand forecasts. In addition to improved estimates of uncertainty of forecast track, other benefits can be derived. For example, probabilistic forecasts of tropical cyclogenesis out to 2 weeks may be realized, and hence information on the likelihood of TC activity can be provided to users. A large ensemble will yield more reliable probability distributions that are often non-Gaussian, such as a bimodal distribution of track depending on the timing of the interaction of a TC and a mid-latitude trough. On a more advanced level, probability distributions of the wind field and rainfall amount would be useful in estimating the risks, and accordingly making decisions on evacuations, closure of businesses and schools, and emergency management preparations. While the wind probabilities issued by the NOAA National Hurricane Center since 2006 are currently based on an average of historical forecast errors, these distributions can in the future be based on a reliable ensemble of the actual forecast. The identification of how initial condition errors and errors in the models contribute to the reduction in forecast skill is also possible with a large, multi-model ensemble, and would in turn lead to improved construction of new ensemble forecasts. An increasingly widespread use of ensembles is to prescribe error covariance statistics in data assimilation, and preliminary research suggests potentially significant advantages over operational data assimilation schemes. The utilization of probabilistic forecast guidance necessitates new collaborative efforts between researchers and forecasters. Among the key areas are: the calibration of the ensemble, the

368

conversion of a discrete ensemble forecast into a continuous probability density function, and the creation of new products that are relevant to the forecaster. The forecaster then has the responsibility to convey the complex information to users. During PARC, researchers will have direct (and for the first time, easy) access to the ensemble forecast outputs from TIGGE and NAEFS. These outputs will be used to provide improved ensemble-based adaptive sampling guidance. TIGGE will also allow for study of whether ensemble-based methods are able to produce improved probabilistic forecasts of high-impact events, as well as providing for new insights into TC structure and dynamics. The TIGGE data sets will allow for easy access of societal and impact researchers to probabilistic data. Basic research and development are required to develop approaches that meet the decision-making needs of specific forecast users and to accurately represent the overall economic benefit. Hydrological Ensemble Prediction Experiment (HEPEX): The hydrological community is making a large effort to utilize ensemble modeling for hydrological applications. PARC focuses on precipitation events over Asia, the Arctic, and North America. An opportunity exists for significant collaboration between PARC, and HEPEX to further the utilization of weather information in ensemble prediction systems for a season. 2b 3.5 Understanding and improving society’s response to weather disasters, including the appropriate use and evaluation of probabilistic information PARC and all THORPEX programs include a strong societal and economic impacts (SEA) subprogram. Specific SEA research opportunities in PARC, related to quantifying the benefits of the program and improving the utilization of weather information include: Forecasting west coast severe weather: PARC investigators will work with several forecast offices on the west coast of the US and Canada to quantify the impacts of PARC research and additional measurements on their forecasts. The work will include gaining insight into the forecast improvements made from PARC measurement strategies, development and testing of conceptual models of the global to regional dynamics of high-impact weather, and improved understanding of the user needs and forecasters view of model short-comings. Heavy rainfall, tropical cyclones, comparative human responses, and hazard mitigation/response: In terms of damage and human misery, typhoons are the overwhelming disaster of developing countries of temperate and tropical eastern Asia. PARC research activities will include: i) Undertake studies to identify populations specifically vulnerable to weather hazards and assess methods for improving forecast, warning, and response systems specific to the needs and capabilities of this/these populations; ii) Estimation of the health impacts of weather of heavy rainfall/typhoons and the degree to which negative outcomes can be mitigated; iii) Decision making in emergency management, how weather information is utilized in the face of major risk, and whether probabilistic information can be used to improve decisions and communication; iv) The different responses of the developing and developed world to significant weather disasters. User-specific forecast use and benefits: Research will investigate the use of weather forecast information and its value in decision-making for specific groups of forecast users in Asia, Western North America and/or other regions affected by PARC forecasts. This will include research on users’ perceptions of and preferences for different types of weather forecast information, the role of different types of forecast information in users’ decisions, and the value of different types of forecast information services. The effort will focus on current and improved short- to mid-term (up to 14 days) weather forecast information related to PARC, including (TIGGE and NAEFS) probabilistic forecasts and other information about forecast uncertainty. Possible user groups to study include households (the public), water managers, energy sector decision makers, and/or emergency managers in developed countries.

369

The intent is to use state-of-the-art survey, and comprehensive multivariate analysis of the results. User-driven verification and evaluation of forecast quality: Traditional forecast verification approaches often do not link forecasts and the decision making process of users, due to their reliance and focus on single-measure scores to summarize forecast performance. 2b 4.0 Experimental Design and Observational Requirements PARC represents a multinational collaboration that incorporates a variety of funding sources. The description of the experiment schedule, design, and data requirements are defined relative to the Asian THORPEX program and the North American THORPEX program. The specific scientific objective(s) satisfied by the deployment of each data facility is defined relative to the objectives defined in Sections 2 and 3. The overall PARC period is scheduled for July-December 2008. During July and August 2008, the Asian THORPEX component of PARC begins with an emphasis on the objectives associated with increased predictability of tropical cyclone formation, track, structure, and landfall. During this period, data facilities will include manned and unmanned aircraft, driftsondes, and satellite data. Operational and experimental forecast products from several national weather centers will provide global ensemble forecasts, control forecasts, and identification of targeted areas for initial condition sensitivities. A portion of the Asian THORPEX program for PARC is comprised of the Dropwindsonde Observations for Typhoon Surveillance near the Taiwan Region (DOTSTAR) program (Wu et al. 2004), which has provided aircraft dropwindsonde observations of tropical cyclones over the Philippine Sea during the past several summer periods. While DOTSTAR is designed to examine tropical cyclones prior to recurvature over the East China Sea, Japan proposes to pursue aircraft dropwindsonde observations during and following recurvature to support sensitivity analysis and a downscaling of global ensemble forecasts of tropical cyclone track, intensity, and structure. These programs will be synchronized with the North American THORPEX PARC objectives for assessment of the impacts of increased predictability associated with the tropical cyclone prior to extratropical transition. The North American THORPEX component of PARC proceeds from September – October 2008 to coincide with the climatological maximum in recurving tropical cyclones that undergo extratropical transition (Jones et al. 2003). The Asian THORPEX component will continue during the September period at a minimum. The North American THORPEX component consists of manned and unmanned aircraft, driftsondes, and satellite data. Because of the complex and varied physical mechanisms associated with extratropical transition and the forcing of eastward-propagating wave packets, which typically occurs over oceanic regions, a mix of aircraft capabilities is required to adequately sample relevant regions of the decaying tropical cyclone and the midlatitude environment into which it is moving. Following the termination of the PARC component dedicated to tropical cyclones and extratropical transitions, the focus turns to extratropical cyclogenesis in the primary Asian wave guide. During November this will primarily be monitored with unmanned aircraft, driftsondes and special satellite observations. During December 2008, it is possible that manned aircraft will be added to this portion of PARC if the NOAA G-IV becomes available. 2b 5.0 Discussion The PARC initiative is an excellent example of a tropical cyclone related research study within THORPEX. The IWTC-VI special session will also include a broad discussion on THORPEX, with input from the research and forecast community on PARC, as well as other potential tropical cyclone related activities within THORPEX.

370

2b 6.0 Bibliography Aberson, S.D., 2001: The Ensemble of Tropical Cyclone Track Forecasting Models in the North Atlantic Basin (1976-2000). Bull. Amer. Meteor. Soc., 82, 1895-1904. Aberson, S. D., 2003: Targeted observations to improve operational tropical cyclone track forecast guidance. Mon. Wea. Rev., 131, 1613-1628. Agusti-Panareda, A., C. D. Thorncroft, G. C. Craig, and S. L. Gray, 2004: The extratropical transition of hurricane Irene (1999): A potential vorticity perspective. Q. J. R. Meteorol. Soc., 130, 1047-1074. Bister, M., and K. A. Emanuel, 1997: The genesis of Hurricane Guillermo: TEXMEX analyses and a modeling study. Mon. Wea. Rev., 125, 2662–2682. Chang, E. K. M., 2005: The impact of wave packets propagating across Asia on Pacific cyclone development. Mon. Wea. Rev., 133, 1998-2015. Chang, E. K. M., and D. B. Yu, 1999: Characteristics of wave packets in the upper troposphere. Part I: Northern Hemisphere winter. J. Atmos. Sci., 56, 1708-1728. Charron, M. and G. Brunet, 1999: Gravity Wave Diagnosis Using Empirical Normal Modes. J. Atmos. Sci., 56, 2706-2727. Danielson, R. E., J. R. Gyakum, and D. N. Straub, 2004: Downstream baroclinic development among forty one cold-season eastern North Pacific cyclones. Atmosphere-Ocean, 42, 235-250. Ferranti, L., T.N. Palmer, F. Molteni and E. Klinker, 1990: Tropical-Extratropical Interaction Associated with the 30-60 Day Oscillation and Its Impact on Medium Range and Extended Range Prediction, J. Atmos. Sci., 47, 2177-2199. Franklin, J.L., C. J. McAdie and M. B. Lawrence. 2003: Trends in Track Forecasting for Tropical Cyclones Threatening the United States, 1970-2001. Bull. Amer. Met. Soc., 84, 1197-1203. Hakim, G. J., 2003: Developing wave packets in the North Pacific storm track. Mon. Wea. Rev., 131, 2824-2837. Hakim, G. J., 2005: Vertical structure of midlatitude analysis and forecast errors. Mon. Wea. Rev., 133, 567-575. Harr, P. A., M. S. Kalafsky, and R. L. Elsberry, 1996a: Environmental conditions prior to formation of a midget tropical cyclone during TCM-93. Mon. Wea. Rev., 124, 1693-1710. Harr, P. A., R. L. Elsberry, and J. C. L. Chan, 1996b: Transformation of a large monsoon depression to a tropical storm during TCM-93. Mon. Wea. Rev., 124, 2625-2643. Harr, P. A., and R. Elsberry, 2000: Extratropical transition of tropical cyclones over the western North Pacific. Part I: Evolution of structural characteristics during the transition process. Mon. Wea. Rev., 128, 2613-2633. Harr, P. A., and R. Elsberry, 2000: Extratropical transition of tropical cyclones over the western North Pacific. Part II: The impact of midlatitude circulation characteristics. Mon. Wea. Rev., 128, 2613-2633. Hendricks, E. A., M. T. Montgomery, and C. A. Davis, 2004: On the role of “vortical” hot towers in tropical cyclone formation. J. Atmos. Sci., 61, 1209–1232.

371

Hoskins, B.J., and Kevin I. Hodges. 2002: New Perspectives on the Northern Hemisphere Winter Storm Tracks. J. Atmos. Sci. 59, 1041–1061. Jones, S. C., et al., 2003: The extratropical transition of tropical cyclones: Forecast challenges, current understanding, and future directions. Wea. Forecasting, 18, 1052-1092. Keen, R. A., 1982: The role of cross-equatorial tropical cyclone pairs in the Southern Oscillation. Mon. Wea. Rev., 110, 1405–1416. Klein, P. M., P. A. Harr, and R. Elsberry, 2000: Extratropical transition of western North Pacific tropical cyclones: An overview and conceptual model of the transformation stage. Wea. Forecasting, 15, 373-395. Klein, P. M., P. A. Harr, and R. Elsberry, 2002: Extratropical transition of western North Pacific tropical cyclones: Midlatitude and tropical cyclone contributions to reintensification. Mon. Wea. Rev., 130, 2240-2259. Lander, M. A., 1990: Evolution of the cloud pattern during the formation of tropical cyclone twins symmetrical with respect to the equator. Mon. Wea. Rev., 118, 1194–1202. Langland, R. 2006: Issues in Targeted Observing. (Invited Paper for Special Symposium Issue). In press, Quart. J. Roy. Meteor. Soc. Majumdar, S. J., C. H. Bishop, B. J. Etherton, and Z. Toth, 2002: Adaptive sampling with the ensemble transform Kalman filter. Part II: Field program implementation. Mon. Wea. Rev., 130, 1144-1165. Majumdar, S. J., S. D. Aberson, C. H. Bishop, R. Buizza, M. S. Peng and C. A. Reynolds, 2006: A comparison of adaptive observing techniques for Atlantic tropical cyclones. Mon. Wea. Rev., in press. McAdie, C.J. and Miles B. Lawrence. 2000: Improvements in Tropical Cyclone Track Forecasting in the Atlantic Basin, 1970-98. Bull. Amer. Meteor. Soc., 81, 989-997. McMurdie, L., and C. Mass, 2004: Major numerical forecast failures over the Northeast Pacific. Wea. Forecasting, 19, 338-356. Montgomery, M.T., and Janice Enagonio. 1998: Tropical Cyclogenesis via Convectively Forced Vortex Rossby Waves in a Three-Dimensional Quasigeostrophic Model. J. Atmos. Sci. 55, 3176 –3207. Montgomery, M. T., M. E. Nicholls, T. A. Cram, and A. B. Saunders, 2006: A vertical hot tower route to tropical cyclogenesis. J. Atmos. Sci., in press. Nielsen-Gammon, J. W., and R. J. Lefevre, 1996: Piecewise tendency diagnosis of dynamical processes governing the development of an upper-tropospheric mobile trough. J. Atmos. Sci., 53, 3120-3142. Orlanski, I., and J. P. Sheldon, 1995: Stages in the energetics of baroclinic systems. Tellus, 47A, 605-628. Parker, D.J. and A. J. Thorpe. 1995: Conditional convective heating in a baroclinic atmosphere: A model of convective frontogenesis. J. Atmos. Sci., 52, 1699–1711. Peng, M. S., and Reynolds, C. A., 2006: Sensitivity of tropical cyclone forecasts.

372

Powell, M.D. and S. D. Aberson. 2001: Accuracy of United States Tropical Cyclone Landfall Forecasts in the Atlantic Basin (1976-2000). Bull. Amer. Meteor. Soc., 82, 2749-2767. Reynolds, C.A. and R. Gelaro, 2001: Remarks on Northern Hemisphere Forecast Error Sensitivity from 1996 to 2000. Mon. Wea. Rev., 129, 8, 2145-2153. Sardeshmukh, P. D., and B. J. Hoskins, 1988: The generation of global rotational flow by steady idealized tropical divergence. J. Atmos. Sci., 45, 1228-1251. Shapiro, M.A., and A.J. Thorpe, 2004: THORPEX International Science Plan, Version 3, 2 November 2004 WMO/TD No. 1246 WWRP/THORPEX No. 2 Sobel, A. H. and S.J. Camargo 2005: Influence of Western North Pacific Tropical Cyclones on Their Large-Scale Environment. J. Atmos. Sci., 62, 3396-3407. Szunyogh, I., Z. Toth, A. V. Zimin, S. Majumdar, and A. Persson, 2002: Propagation of the effect of targeted observations: The 2000 Winter Storm Reconnaissance Program. Mon. Wea. Rev., 130, 1144-1165. Wu, C.-C., P.-H. Lin, S. D. Aberson, T.-C. Yeh, W.-P. Huang, J. -S. Hong, G.-C. Kiu, K.-C. Hsu, I.-I. Lin, K.-H. Chou, P.-L. Lin, and C.-H. Liu, 2004: Dropsonde observations for typhoon surveillance near the Taiwan region (DOTSTAR): An overview. Bull. Amer. Meteor. Soc., 86, 878-790. Wu, C.-C., P.-H. Lin, J.-H. Chen, and K.-H. Chou, 2005: Targeted observations of tropical cyclones based on an adjoint sensitivity steering vector. Geophys. Res. Letters, submitted.

373

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES Topic 3 : TROPICAL CYCLONE MOTION Topic Chair: Russell L. Elsberry Graduate School of Engineering and Applied Sciences Department of Meteorology Naval Postgraduate School Monterey, California 93943 USA

Email: elsberry@nps.edu Fax: 831-655-3061 3.0.1 Introduction Already at IWTC-V, this topic was focused on tropical cyclone (TC) track prediction rather than on observational, theoretical, or numerical simulations of TC motion, as had been the case in the early IWTCs. The focus in this report will again be on TC track prediction rather than TC motion. One of the highlights of Topic 3 at IWTC-V was the report on operational consensus track forecasting at the Joint Typhoon Warning Center (JTWC) that had resulted in dramatic improvements in 72-h track forecast accuracy in the western North Pacific (Jeffries and Fukada 2002). R. Jeffries also presented a special focus session on consensus forecasting that was well attended. Jeffries and Fukada (2002) also presented the internal tests of 96-h and 120-h track forecasts at JTWC. Based on the success of these tests, and similar testing at the U. S. National Hurricane Center, both warning centers announced their plans to begin issuing official forecasts to 120 h during the 2003 season. 3.0.2 Advances in operational track prediction The improvements in track prediction are often attributed to better guidance from Numerical Weather Prediction (NWP) models. Improved NWP track guidance by the Japan Meteorological Agency (JMA) and the United Kingdom Meteorological Office (UKMO)is documented in Topic 3.1. In addition, the forecasters are making better use of the NWP guidance through multiple-model consensus forecasting, as the consensus track errors over a season are smaller than any of the individual model errors. Ensemble prediction system (EPS) track forecasts are also becoming available in some warning centers. Some post-processing of the EPS output has improved the strike probability maps generated from the European Center for Medium-range Forecasts (ECMWF)EPS, which includes 50 members. The improvements in 24-, 48-, and 72-h TC track forecasts by various warning centers and for various ocean basins are then documented in Topic 3.1. Particularly dramatic improvements in 24-h and 48-h track forecasts have been achieved in the North Indian Ocean and the Southern Hemisphere from the 1990s to the present. Long-term trends in the 72-h errors may only be evaluated for the U.S. warning centers since the only non-U.S. center to make 72-h forecasts is the JMA and they have only been issuing 72-h track forecasts since 2001. Particularly large improvements are again noted in the 72-h forecasts for the North Indian Ocean and the Southern Hemisphere. The differences among the basins are now much smaller than during the 1990s, and the remaining differences are probably related to the degree of difficulty in the basins, and/or small sample sizes.

374

In summary, several operational warning centers in the developed nations have made use of improved NWP track guidance to make more accurate track forecasts through 72 h in all ocean basins. The 120-h track forecast accuracy for the U.S. warning centers is summarized in Topic 3.2. The 120-h errors are generally about 300 n mi, which was a typical value for 72-h errors in 1990. This is another dramatic improvement in tropical cyclone track forecasting, which is attributed to better NWP guidance and better use of this guidance via consensus forecasting. Given the importance of the NWP guidance, Topic 3.2 includes a summary of recent upgrades in the numerical models in many countries. Because of the importance of consensus track forecasting, a special focus session 3a on this topic will be offered at IWTC-VI to share experiences. Both weighted and non-weighted methods of combining multiple NWP model tracks will be described. A selective consensus approach at JTWC for the 72-h forecasts described at IWTC-V (Jeffries and Fukada 2002) has been dropped in favor of a non-selective 10-member consensus. However, only four NWP model tracks are available at JTWC for 96-h and 120-h forecasting. When one or more of these NWP models has a significant error, the selective consensus of the remaining models may be more accurate. Payne et al. (2006) indicate that the average improvement in the 120-h selective consensus forecasts relative to the non-selective consensus forecasts would be 222 (239) n mi during 2005 (2004), and the corresponding average improvement relative to the JTWC forecasts would be 282 (203) n mi. Even though these selective consensus forecasts would have been appropriate for only 20-25% of the 120-h forecasts, their proper formulation would have significantly improved the seasonal error statistics. A new method by Jim Goerss of the Naval Research Laboratory-Monterey for determining the track forecast confidence based on the spread of the dynamical model guidance will also be described in the special focus session 3a. Whenever the spread is small, a smaller circle can be drawn around the consensus mean position to indicate the likelihood that about 72-74% of the time the storm center will be inside the circle. This representation of the track forecast confidence takes into account the degree of difficulty of the forecast, as opposed to drawing the same confidence circle in every situation. The NWP model guidance is not always accurate at 120 h. Since many of the verifiable 120-h forecasts involve an interaction with the midlatitude circulations, a large fraction of the 120-h errors arise from an improper prediction of this interaction. For example, Kehoe et al. (2006) found 83% (85%) of the U. S. Navy global (regional) model errors in the western North Pacific during the 2004 season were due to midlatitude errors. Similarly, about 90% of all large (> 500 n mi at 120 h) errors by these two models and the NCEP and UKMO global models during the 2005 western North Pacific season were attributed to midlatitude-related sources (Payne et al. 2006). Frequent, systematic errors in the Navy regional model and the NCEP global model could have been recognized from examining the fields. A further description of the various EPS that might be used to extract tropical cyclone tracks is given in Topic 3.2. These EPSs may involve a single model with perturbed initial conditions for each member, or a multiple model variation may also be included. In the near future, a THORPEX Interactive Grand Global Ensemble (TIGGE) that may include the combined EPSs of perhaps 11 countries will become available. Tropical cyclone tracks might be requested as an output from the TIGGE since this might give an estimate of the track uncertainty. 3.0.3 Data assimilation Plans at seven NWP centers for data assimilation of satellite data were presented at IWTC-V. While these data assimilation systems were not specifically for tropical cyclones, Topic 3.2 also describes some recent three-dimensional variational and four-dimensional variational systems and Ensemble Kalman Filter systems that may have specific applications for tropical cyclone prediction.

375

Some data assimilation issues are also discussed in Topic 3.3, and especially the need for flow-dependent background error covariance in the region of the tropical cyclone. When targeted observations from aircraft (radar data as well as dropwindsondes) and other special satellite instruments are available, special data assimilation considerations will be necessary. The data assimilation system can be used to determine which observations have the greatest contribution to the tropical cyclone track forecast. 3.0.4 Targeted observations Recent progress in the use of targeted observations to improve tropical cyclone track prediction is summarized in Topic 3.3. Whereas the capability to deploy dropsondes in the environment of tropical cyclones had previously been limited to the U.S., the Dropwindsonde Observations for Typhoon Surveillance near the Taiwan Region (DOTSTAR) program has opened the possibility of targeted observation. Four techniques for determining the most sensitive regions at the observation time are described in Topic 3.3. Sometimes the sensitive areas are in the region of the tropical cyclone and at other times the sensitive areas are remote from the cyclone. Whereas some preliminary comparisons of the forecast errors for analyses based on different targeted techniques are presented, it appears that larger sample sizes are required for conclusive results as to the superior technique. 3.0.5 Recommendations Whereas multiple recommendations are given in Topics 3.1 and 3a, the consistency between these recommendations is noteworthy. Because of the success in reducing track forecast errors by the operational centers in developed countries, these recommendations are intended to provide the tracks, tools, and training to warning centers in other countries. Given the improvements in track forecasting described in Topic 3.1, it can be confidently expected that similar improvements would also be achieved in other countries. Ensemble prediction systems continue to be developed that have potential for tropical cyclone track prediction. Research and training on how to most effectively use these ESP tracks is required. A recommendation to continue and expand the targeted observation programs would be appropriate. 3.0.6 References Jeffries, R. A., and E. J. Fukada, 2002: Consensus approach to track forecasting. Paper TP3.2, Extended abstracts, Fifth International Workshop on Tropical Cyclones, Cairns, Australia, World Meteorological Organization (Geneva). Kehoe, R. M., M. A. Boothe, and R. L. Elsberry, 2006: Dynamical tropical cyclone 96-h and 120-h track forecast errors in the western North Pacific. Weather and Forecasting, (in revision). Payne, K. A., R. L. Elsberry, and M. A. Boothe, 2006: Assessment of western North Pacific 96-h and 120-h track guidance and present forecastability. Weather and Forecasting, (in review).

376

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES

Topic 3.1 : Advances and Requirements for Operational Tropical Cyclone Track Prediction Rapporteur: T.C. Lee Hong Kong Observatory 134A Nathan Road, Kowloon, Hong Kong, China. Email : tclee@hko.gov.hk Fax : (852) 2377 3472 Working Group: Wes Browning, Philippe Caroff, Eun-Jeong Cha, Duong Lien Chau, Cuiying Tian, Julian Heming and Alan Radford, Prisco Nilo, Mannoji Nobutak, David Richardson, Ali Shareef, S.K. Subramanian. 3.1.1. Introduction A National Meteorological and Hydrological Service (NMHS) should provide accurate quantitative forecasts of high winds, heavy rain, and storm surge associated with tropical cyclones. Under-warnings of these hazards may result in avoidable loss of lives and damage to properties while over-warnings could cause unnecessary disruption to socio-economic activities. Both affect the credibility of the NMHS. An accurate forecast of tropical cyclone (TC) track and a good knowledge of the wind and rain distribution around the TC are essential in this regard. In this report, the recent developments in operational TC track prediction techniques and tools (Section 3.1.2) and the improvement in operational TC track forecasts (Section 3.1.3) in the past few years are reviewed. Identified roadblocks in operational TC track prediction are discussed in Section 3.1.4. Opportunities and recommendations are given in Section 3.1.5. In addition to the inputs of working group members, this report draws on published information as well as TC track forecast verification data available online from the websites of various operational warning centres. 3.1.2. Recent developments in operational tropical cyclone track prediction techniques and tools (a) Improvements in numerical weather prediction (NWP) guidance With increasing horizontal and vertical resolution, improvement in physical parameterization schemes, and more effective assimilation of satellite observations, the skill of numerical weather prediction (NWP) models in TC track prediction has been improved significantly in the last decade (Heming 2000; Lam 2001; Aberson 2001; Sakai and Yamaguchi 2004; Goerss and Sampson 2004). Some examples are the 5-year mean TC track forecast errors of the Japan Meteorological Agency (JMA) global model for western North Pacific and South China Sea from 1995 to 2005 (Fig. 3.1.1a), 5-year mean TC track forecast errors of the UK Meteorological Office (UKMO) global model for Northern Hemisphere from 1992 to 2005 (Fig. 3.1.1b), and 5-year mean TC track forecast errors of UKMO for Southern Hemisphere from 1992 to 2005 (Fig. 3.1.1c). The forecast errors of these two models for all forecast ranges from 24 hours to 120 hours have decreased over the last decade, with the most dramatic improvements in the longer forecast ranges. The 5-year mean errors of 72-hour forecasts for 2001-2005 are now smaller than those of 48-hour forecasts for 1991-1995. The reductions during 1996-2000 and 2001-2005 were in the range of 10-15 % for 24- and

377

48-hour forecasts (see Table 3.1.2). Moreover, NWP models are now capable of even predicting sharp turns in the TC tracks. A recent example (Fig. 3.1.2) is the forecast for Typhoon Chanchu in May 2006 that abruptly made a right-angle turn in the South China Sea. (b) Enhancement in availability and accessibility of NWP products and operational TC forecast

information More NWP products are made easily accessible for NMHSs via either the WMO Global Telecommunication System (GTS) or internet through bilateral or regional arrangements. For example, JMA implemented in 2004 a password-protected "Numerical Tropical Cyclone Prediction Web Site" that displays TC forecast tracks from major NWP centres for access by the Members of the ESCAP/WMO Typhoon Committee (Kyouda 2006). Furthermore, official TC warnings issued by Regional Specialized Meteorological Centres (RSMCs) and NMHSs are available at the Severe Weather Information Centre (SWIC) website (http://severe.worldweather.wmo.int/) of the WMO (Lam 2001). (c) Developments in multi-model consensus TC track forecasts Goerss (2000), Lee and Wong (2002), and Jeffries and Fukada (2002) indicate that the TC forecast tracks derived from the multi-model consensus are, on average, more accurate than the forecasts of individual models. The multi-model consensus approach that makes simple averages of the forecast tracks from different centres is a relatively low cost solution for adding values to model TC track forecasts and has been adopted by several warning centres as the primary guidance for TC track forecasts in recent years. Besides simple averaging, other methods such as selective consensus method(Carr et al. 2001), weighted consensus(Weber 2003), and statistical/linear regression schemes (Vijaya Kumar et al. 2003; Zhang 2006) have been developed to reduce the error of the consensus track. Furthermore, a consensus prediction error product has also been developed to provide a measure of confidence in the consensus forecasts (Goerss 2004, 2006). These radii of circular areas indicate the uncertainty at a certain confidence level of the forecast TC positions based on the spread of the consensus each day, rather than being the same radii each day based on historical average errors. In recent years, the outputs of Ensemble Prediction Systems (EPS) of major NWP centres have been made available to NMHSs (Cheung 2001; Heming et al. 2004; Kyouda 2006). From individual TC track forecasts of EPS, TC strike probability forecasts can be generated for use by forecasters and decision-makers (Regnier and Harr 2006). Two examples of strike probability maps generated by the European Centre for Medium-range Forecasts (ECMWF) ensemble for Hurricane Katrina are given in Fig. 3.1.3. The ensemble track forecasts from 00 UTC 26 August suggest a large uncertainty in the track of Katrina. Once Katrina crossed into the Gulf of Mexico 12 hours later, the uncertainty in the track of Katrina was substantially reduced. The ECMWF strike probability maps are available to all NHMSs via the ECMWF website. As recommended by the IWTC-V, the TC tracks for the individual ensemble members are now distributed on the GTS. One example of how EPS outputs may be used to assist forecasters is in handling bifurcation situations. Under a cooperative research project between HKO and JMA on the utilization of the EPS TC data, Wong (2006) found that the use of actual TC observed positions a posteriori for the “conditioning” of strike probability could help forecasters to choose among divergent EPS tracks such as in the case of Songda (Fig. 3.1.4). As verified by the actual track (black line in Fig. 3.1.4), the conditioned strike probability successfully indicated the likelihood of Songda recurving near 30o N, 128o E. (d) Development of tools for operational TC track forecasting Several NMHSs have developed integrated, interactive tools to assimilate and display track forecast information and to interpolate model tracks to facilitate the formulation of forecasts and warning strategies by front-line forecasters. Figure 3.1.2 is an example of the Tropical Cyclone Information Processing System (TIPS) operated in Hong Kong, China (Tai and Ginn 2001).

378

3.1.3. Improvements in operational tropical cyclone track forecasts The accuracy of 24-, 48-, and 72-h track forecasts for eight centres from 1990 to 2005 using data provided by working group members and available online (see Table 3.1.1) was reviewed (Figs. 3.1.5a-c). These centres are: Regional Specialized Meteorological Centre-Tokyo (RSMC-Tokyo); Regional Specialized Meteorological Centre-La Reunion (RSMC-La Reunion); National Weather Centre of China Meteorological Administration (CMA); Joint Typhoon Warning Center of U.S. Department of Defense (JTWC); National Hurricane Center of U.S.A. (NHC); Central Pacific Hurricane Center of U.S.A. (CPHC); Vietnam National Center for HydroMeteorological Forecasting (NCHMF); and Hong Kong Observatory (HKO). A 5-year running mean of the track forecast errors has been used to smooth the interannual variations in track errors and the number of TC forecasts issued by various centers and thus illustrate the long-term trend. Significant improvements have been achieved by practically all of the centres. In general, the 24-, 48- and 72-hour TC track forecast errors have been reduced to around 150 km, 250 km, and 350 km, respectively, during 2001-2005. Except for CPHC, the reductions in the 5-year mean 24- and 48-hour TC track forecast errors in the two consecutive 5-year periods of 1996-2000 and 2001-2005 were in the range of 10% to 35%. These reductions are commensurate with the trends in the track forecast errors of the JMA global model and the UKMO global model for the same two 5-year periods. Moreover, reductions in errors for some NMHSs and RSMCs are noticeably higher than the 10% to 15% of individual models, which is attributed to the use of the multi-model consensus and the EPS products value to the model TC track forecasts (Goerss 2000; Lee and Wong 2002; Jeffries and Fukada 2002; Goerss et al. 2004; Heming et al. 2006). 3.1.4. Roadblocks In spite of the significant developments and advancements as described in sections 3.1.2 and 3.1.3, a number of roadblocks need to be overcome to improve operational TC track forecasting. 3.1.4.1 Lack of observations Although the passive microwave and scatterometer data from polar-orbiting satellites (e.g., AMSU, QuikSCAT, SSM/I, etc) in recent years have been made available for better determination of TC position and intensity, the temporal coverage of these satellites still does not meet all of the needs for real-time operations. The lack of observations for determining the position and intensity of a TC accurately over data-sparse areas (especially for weaker TCs with ill-defined structure) still hinders the numerical and operational TC track prediction. 3.1.4.2 Diversity of NWP guidance in forecasting TC track While the multi-model consensus provides reliable guidance for formulation of TC track forecasts in the majority of cases when the model tracks are in reasonably good agreement, in other occasions the models give very diverse forecasts. This diversity might be due to uncertainties in initial conditions for weaker storms, difficulties in predicting certain synoptic patterns, or model bias. Simple consensus forecasts usually do not work in such cases. While the selective consensus method (Carr et al. 2001) might be used to identify and remove the likely erroneous model or cluster, very large forecast errors could still occur if the wrong model track is removed. These bifurcation cases therefore pose a great challenge to forecasters (Carr and Elsberry 2000; Jeffries and Fukada 2002). 3.1.4.3 Difficulty in forecasting landfalling TCs It makes a lot of difference to disaster preparedness and prevention for a particular location if the forecast track of a TC is off by tens of kilometres. Current NWP guidance with an average 24-hour track forecast error of about 150 km is not sufficiently accurate for the protection of a city (Lee 2002; Regnier and Harr

379

2006). More accurate forecasts of the landfall point of a TC as well as the occurrence of gale winds are required. 3.1.4.4 Limited accessibility of NWP products and lack of operational TC track forecasting tools The TC track forecasts from major NWP centres are not available to all NMHSs. Some NMHSs only have access to the outputs of no more than one or two NWP models. Multi-model consensus tracks that provide better accuracy than individual model forecasts thus cannot be constructed. Those NMHSs already using multi-model consensus method require more NWP products to identify the optimal combination of forecast members (Goerss and Sampson 2004). Some NMHSs also lack resources and experience to develop interactive graphic tools for assimilating, interpolating, and displaying TC forecast information from advanced centres in an operational environment. 3.1.4.5 Lack of training Many forecasters and meteorologists lack the appropriate training to make good use of NWP products from advanced centres. 3.1.5. Opportunities and recommendations 3.1.5.1 More observations for determining TC position and intensity The National Polar-orbiting Operational Environmental Satellite System (NPOESS) satellites are scheduled to be launched in the next few years. This will greatly add to the satellite-derived wind data that can be ingested in NWP models (Velden and Hawkins 2002; Miller et al. 2006). Thus, NWP centres should prepare for the ingestion of these satellite data into their numerical models. The NMHSs are encouraged to disseminate their weather observation data and radar fix (RADOB) data via GTS or internet as much as possible when a TC is within the coverage of their observation networks. 3.1.5.2 Development and utilization of EPS products In recent years, post-processing of EPS TC track forecast has been explored by several NMHSs to improve TC track prediction. It is recommended that further studies in this area should be conducted with a view to developing a systematic and optimized approach for operational implementation. The consensus prediction error product (Goerss 2004, 2006) and outputs of EPS as mentioned in Section 3.1.2(c) can provide forecasters with some measure of the confidence in the TC track forecasts. This confidence measure may also help decision-makers to balance the lead-time versus accuracy trade-off and evaluate investments to reduce lead times or create flexibility in preparations (Regnier and Harr 2006). It is recommended that NMHSs should explore the possibility of integrating the spread information from consensus or the EPS track forecasts with the operational forecast track to also indicate the confidence. 3.1.5.3 Development of mesoscale modeling techniques High-resolution mesoscale models incorporating local climate features and topography hold promise for improving TC track forecasts near landfall. The UKMO is applying an experimental North American (NA) regional model to investigate the impact of resolution on the forecasts for severe weather events such as hurricanes. This NA model has a domain covering North America, the Gulf of Mexico, and Caribbean Sea and has a horizontal resolution of 0.15o lat. x 0.15o long. (Heming et al. 2006). To meet the need for accurate track and wind/rainfall forecasts for landfalling TCs, the China Meteorological Administration has been developing a new mesoscale model called GRAPES (Global/Regional Assimilation PrEdication

380

System) since 2001 (Chen and Xue 2006). A version of GRAPES with a resolution of 0.25o lat. x 0.25o

long. was experimentally run in 2004 and its outputs have been used for TC track and precipitation forecasts (Fig. 3.1.6). Continuing research efforts are recommended to improve the ability of mesoscale models in simulating the TC landfalling processes. 3.1.5.4 Availability of tropical cyclone forecast information and tools The JMA "Numerical Tropical Cyclone Prediction Web Site" that displays in graphical form TC forecast tracks provided by major NWP centres is extremely useful to Typhoon Committee Members (RSMC-Tokyo 2003; Kyouda 2006). This website is an encouraging example for sharing NWP products among NMHSs. It is recommended that similar initiatives be developed for other TC basins and related forecasting tools and references be made available to NMHSs. To facilitate timely and convenient access to NWP products by NMHSs, it is recommended that the NWP centres disseminate TC track forecasts and EPS products in a standardized digital / alphanumeric format via GTS and internet. The NWP and operational TC track forecasts developed in the past few years are very useful for NMHSs to identify optimal combinations of forecast tracks for the multi-model consensus and for researchers to develop new TC track forecasting techniques and tools. It is recommended that a feasibility study be made for setting up a unified database for archiving and sharing of NWP and operational TC track and intensity forecasts. 3.1.5.5 Collaboration and Training It is recommended that more training events for forecasters be organized on the use of multi-model consensus and EPS products in operational TC track forecasting. Opportunities should be created for interaction between researchers and forecasters, such as through regional or international meetings, seminars and workshops on Operational Tropical Cyclone Forecasting. These events would enable: (i) NMHSs to provide feedback on the usefulness of the NWP models; (ii) forecasters to share experiences on operational TC forecasting (e.g. forecasting tools, procedures, etc.) and thus help model developers and researchers to better understand the needs and difficulties of the NMHSs; and (iii) the identification of research needs and how research results could be adopted for operational use. 3.1.6 Summary Considerable progress has been made in operational TC track forecasting in the past decade. The 5-year mean of the 24-, 48- and 72-hour forecast errors of the eight RSMCs/NMHSs studied have been reduced to around 150 km, 250 km, and 350 km, respectively, during 2001-2005, roughly 10% to 35% less than those during 1996-2000. Such improvements are attributed to the improved TC track forecasting guidance from NWP models and the use of the multi-model consensus forecasts and EPS products. However, more reliable objective forecasting techniques have yet to be developed to handle divergent track scenarios and to improve short-range forecasting of the landfall point. In this respect, methods for further post-processing of EPS track forecasts have to be developed. More TC track guidance from NWP centres should be made available to forecasters. Steps should be taken to ensure timely and convenient access to NWP model forecast tracks in a standardized format by all NMHSs. Moreover, forecasters in many NMHSs require software tools and appropriate training to fully utilize the improved NWP-based TC forecast information in the operational environment.

381

References Aberson, S. D., 2001: The ensemble of tropical cyclone track forecasting models in the North Atlantic Basin (1976-2000). Bull. Amer. Meteor. Soc., 82 , 1895-1904. Bell, G. J., 1979: Operational forecasting of tropical cyclones: past, present and future. Aust. Meteor. Mag., 27, 249-258. Carr, L. E., III, and R. L. Elsberry, 2001: Beta test of the systematic approach expert system prototype as a tropical cyclone track forecasting aid. Wea. Forecasting, 16, 355-368. Carr, L. E., III, and R. L. Elsberry, 2000: Dynamical tropical cyclone track forecasting errors. Part I: Tropical region error sources. Wea. Forecasting, 15, 641-661. Chen, D., and J. Xue, 2006: Introduction of Chinese new generation of numerical prediction model system: GRAPES. Internal communication of China Meteorological Administration. Cheung, K. K. W., 2001: A review of ensemble forecasting techniques with a focus on tropical cyclone forecasting. Meteorol. Appl. 8, 315-332. Goerss, J. S., 2000: Tropical cyclone track forecasts using an ensemble of dynamical models. Mon. Wea. Rev., 128, 1187-1193. Goerss, J. S., 2004: Estimation of tropical cyclone track forecast uncertainty. Preprints, 26th Conference on Hurricanes and Tropical Meteorology, Miami, FL, Amer. Meteor. Soc., 152-153. Goerss, J. S. and C. R. Sampson, 2004: A history of western North Pacific tropical cyclone track forecast skill. Weather and Forecasting, 19, 633-638. Goerss, J. S., 2006: Prediction of tropical cyclone track forecast error for Hurricane Katrina, Rita and Wilma. 27th Conference on Hurricanes and Tropical Meteorology, Monterey, CA, Amer. Meteor. Soc. Goerss, J.S., and T. F. Hogan, 2006: Impact of satellite observations and forecast model improvements on tropical cyclone track forecasts. 27th Conference on Hurricanes and Tropical Meteorology, Monterey, CA, 23-28 April 2006. Heming, J., 2000: Tropical cyclone forecasting – the last decade. NWP Gazette, June, 4-5. Heming, J.T., S. Robinson, C. Woolcock, and K. Mylne, 2004: Tropical cyclone ensemble product development and verification at the Met Office. 26th Conference on Hurricanes and Tropical Meteorology, Miami Beach, FL, Amer. Meteor. Soc., 158-159. Heming, J.T., and G. Greed, 2006: Katrina, Rita and Wilma: Met Office model forecasts. 27th Conference on Hurricanes and Tropical Meteorology, Monterey, CA, Amer. Meteor. Soc. Jeffries, R.A., and E. J. Fukada, 2002: Consensus approach to track forecasting. Proceedings of the Fifth WMO International Workshop on Tropical Cyclone (Topic 3.2), Cairns, Australia 3-12 December 2002. Kyouda, M., 2006: Report on applications of EPS for severe weather forecasting. WMO Commission of Basic Systems, OPAG DPFS, Expert meeting on Ensemble Prediction Systems, Exeter,UK. 6-10 February 2006. Lam, C.C., 2001: Performance of the ECMWF model in forecasting the tracks of tropical cyclones in the

382

South China Sea and parts of the western North Pacific. Meteorological Applications, 8, 339-344. Lam, C. Y., 2001: Tropical cyclone forecasting: Linkage between theory and practice. WMO Workshop on Typhoon Forecasting Research, Jeju,Korea, 25-28 September 2001. Lam, H. K., 2001: The World Meteorological Organization pilot websites: “Severe Weather Information Centre” and “World Weather Information Service,” 34th Session of the ESCAP/WMO Typhoon Committee. Lee, T. C., and M. S. Wong, 2002: The use of multiple-model ensemble techniques for tropical cyclone track forecast at the Hong Kong Observatory. WMO Technical Conference on Data Processing and Forecasting Systems, Cairns, Australia, 2-3 Dec. 2002. Lee, T.C., 2002: Effective warning. Proceedings of the Fifth WMO International Workshop on Tropical Cyclones (Topic 5.3), Cairns, Australia, 3-12 December 2002. Miller, S.D., J. D. Hawkins, J. Kent, F. J. Turk, T. F. Lee, A. P. Kuciaskas, K. Richardson, R. Wade, and C. Hoffman, 2006: NextSat: Previewing NPOESS/VIIRS imagery capabilities. Bull. Amer. Meteor. Soc., 87, 433-446. Regnier, E. and P. Harr, 2006: Information forecasting for hurricane preparation. 27th Conference on Hurricanes and Tropical Meteorology, Monterey, CA, Amer. Meteor. Soc. Weber, H.C., 2003: Hurricane track prediction using a statistical ensemble of numerical models. Monthly Weather Review, 131, 749-769. Sakai, R., and M. Yamaguchi, 2005: The WGNE Intercomparison of tropical cyclone track forecasts by operational global models. WGNE “Blue Book”, available at http://www.cmc.ec.gc.ca/rpn/wgne/ Tai, S.C., and W. L. Ginn, 2001: Tropical cyclone processing systems (TIPS) of the Hong Kong Observatory, Eighth Workshop on Meteorological Operational Systems, ECMWF, 12-16 Nov 2001. Tsuyuki, T., R. Sakai, and H. Mino, 2002: The WGNE intercomparison typhoon track forecasts from operational global models for 1991-2000. WMO Bulletin, 51, 253-257. Velden, C.S., and J. D. Hawkins, 2002: The increasing role of weather satellites in tropical cyclone analysis and forecasting. Proceedings of the Fifth WMO International Workshop on Tropical Cyclones (Topic 0.3), Cairns, Australia, 3-12 December 2002. Kumar, T. S. V., T. N. Krishnamurti, M. Fiorino, and M. Nagata, 2003: Multimodel superensemble forecasting of tropical cyclones in the Pacific. Mon. Wea. Rev., 131, pp. 574–583. WMO, 2005: Thirty-first status report on the implementation of the WMO Tropical Cyclone Programme. Wong, M.C., 2006: Weather-related disaster risks and risk reduction. WWRP/THORPEX Scientific Conference on "Improving the Global Predictability of High Impact Weather including a review of Southern Hemisphere Plans,” Capetown, South Africa, 13-15 February 2006. Zhang, S. F., S. Z. Gao, and Y. Li, 2005: Application of consensus method on the forecast of tropical cyclones in 2004. Internal communication of China Meteorological Administration.

383

Table 3.1.1 Summary of operational TC track forecasts by warning centres, in various basins, and periods reviewed in this report

Warning Centre TC Basin Period

Joint Typhoon Warning Center of U.S. Department of

Defense (JTWC)

Western North Pacific & South China Sea, North Indian Ocean

(NIO), and Southern Hemisphere (SH)

1985-2005

National Hurricane Center of USA (NHC) Eastern Pacific (EPC)

Atlantic

1989-2005

Central Pacific Hurricane Center of USA (CPHC) Central Pacific 1990-2005

National Weather Centre of China Meteorological

Administration (CMA)

Western North Pacific & South China Sea 1991-2005

Regional Specialized Meteorological Centre – Tokyo

(RSMC-Tokyo)

Western North Pacific & South China Sea 1996-2005

Regional Specialized Meteorological Centre- La

Reunion (RSMC- La Reunion)

for TCs with Dvorak current intensity >=3.0

Southwest Indian Ocean 1990/91-2004/05

Hong Kong Observatory (HKO) Western North Pacific & South China Sea 1985-2005

Vietnam National Center for HydroMeteorological

Forecasting (NCHMF)

South China Sea 1996-2005

384

Table 3.1.2 Summary of 5-year mean TC track forecast errors (km) for 1996-2000 and 2001-2005 (Total number of forecasts in brackets) 24-hour Forecast 48-hour Forecast

Centres 5-Year Mean

(1996-2000)

5-Year Mean

(2001-2005)

Improvement

(%)

5-Year Mean

(1996-2000)

5-Year Mean

(2001-2005)

Improvement

(%)

HKO 182 (514) 131 (617) 28 356 (334) 232 (416) 35

CMA 165 (1040) 133 (1908) 20 318 (802) 228 (1551) 28

RSMC-Tokyo 154 (1839) 128 (2289) 17 288 (1394) 232 (1820) 19

NHC-EPC 131 (1227) 111 (1152) 15 240 (940) 190 (877) 21

NHC-Atlantic 144 (1405) 119 (1743) 17 258 (1168) 219 (1410) 15

JTWC-WNP 184 (3174) 128 (3161) 31 321 (2513) 220 (2597) 31

JTWC-NIO 208 (285) 139 (228) 33 380 (186) 252 (168) 34

JTWC-SH 196 (1737) 144 (1028) 27 382 (1414) 251 (810) 34

CPHC 135 (135) 140 (173) -4 273 (93) 251 (117) 8

NCHMF 169 (289) 143 (232) 15 --- --- ---

RSMC-La

Reunion

172 (689) 155 (848) 10 332 (516) 291 (836) 12

NWP Models

UKMO-NH* 146 (2080) 132 (2431) 9 254 (1556) 227 (1886) 11

UKMO-SH** 178 (1125) 161 (767) 9 301 (831) 267 (568) 11

JMA# 169 (633) 145 (740) 14 268 (515) 229 (611) 15 * forecasts for TCs in Northern Hemisphere / ** forecasts for TCs in Southern Hemisphere # forecasts for TCs in western North Pacific and South China Sea Percentage of improvement = 100% x (Mean Error (1996-2000) – Mean Error (2001-2005)) / Mean Error (1996-2000)

385

(Note : Colour version of the Figures are available on the CD version)

Figure 3.1.1(a) 5-year running mean tropical cyclone forecast errors of JMA global model for western NorthPacific and South China Sea from 1995 to 2005 (Data Source : RSMC-Tokyo)

0

100

200

300

400

500

600

700

800

900

1000

1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Year

5-ye

ar m

ean

erro

r (km

)

24-hr F/C48-hr F/C72-hr F/C96-hr F/C120-hr F/C

Figure 3.1.1(b) 5-year running mean tropical cyclone track forecast errors of UKMO global model forNorthern Henisphere from 1992 to 2005 (Data Source : UKMO)

100

200

300

400

500

600

700

800

900

1000

1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Year

5-ye

ar m

ean

erro

r (km

)

24-hr F/C48-hr F/C72-hr F/C96-hr F/C120-hr

386

Fig. 3.1.2 The HKO display of the forecast tracks of Typhoon Chanchu based on NWP model forecasts from 00 UTC on 12 May 2006. The observed track of Typhoon Chanchu is also plotted (in black) for reference.

Figure 3.1.1(c) 5-year running mean tropical cyclone track forecast errors of UKMO global model forSouthern Hemisphere from 1992/3 to 2004/5 (Data Source : UKMO)

0

200

400

600

800

1000

1200

1992-3

1993-4

1994-5

1995-6

1996-7

1997-8

1998-9

1999-2000

2000-1

2001-2

2002-3

2003-4

2004-5

Year

5-ye

ar m

ean

erro

r (km

)

24-hr F/C48-hr F/C72-hr F/C96-hr F/C120-hr F/C

387

Fig. 3.1.3 ECMWF strike probability maps for Hurricane Katrina for EPS forecasts initialized on 26 August at 00 UTC (left) and 12 UTC (right).

Fig. 3.1.4 Conditioned and unconditioned (inset) strike probability maps for Typhoon Songda in September 2004 (Wong 2006)..

388

Figure 3.1.5(a) 5-year running mean error of 24-hour tropical cyclone track forecasts by different operationaltropical cyclone warning centres (1990-2005)

100

150

200

250

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Year

5-ye

ar ru

nnin

g m

ean

erro

r (km

)

HKOCMARSMC-TokyoNHC-EPCNHC-AtlanticJTWC-WNPJTWC-NIOJTWC-SHRSMC-La ReunionNCHMFCPHC

Figure 3.1.5(b) 5-year running mean error of 48-hour tropical cyclone track forecasts by different operationaltropical cyclone warning centres (1990-2005)

150

200

250

300

350

400

450

500

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Year

5-ye

ar ru

nnin

g m

ean

erro

r (km

)

HKOCMARSMC-TokyoNHC-EPCNHC-AtlanticJTWC-WNPJTWC-NIOJTWC-SHRSMC-La ReunionCPHC

389

Fig. 3.1.6 Track (left) and precipitation (right) forecasts for Typhoon Chanchu in May 2006 by the GRAPES, of CMA.

Figure 3.1.5(c) 5-year running mean error of 72-hour tropical cyclone track forecasts by different operationaltropical cyclone warning centres (1990-2005)

250

300

350

400

450

500

550

600

650

700

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Year

5-ye

ar ru

nnin

g m

ean

erro

r (km

)

RSMC-TokyoNHC-EPCNHC-AtlanticJTWC-WNPJTWC-NIOJTWC-SHCPHC

390

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES Topic 3.2 : Improvements in understanding and prediction of Tropical Cyclone (TC) motion. Rapporteur: Sim D. Aberson, NOAA/AOML/Hurricane Research Division, Miami, FL, USA. Email: sim.aberson@noaa.gov. Working Group : Morris Bender, Roberto Buizza, James Goerss, Gregory Hakim, T. N. Krishnamurti, Martin Leutbecher, Sharanya Majumdar, Steve Tracton, Gregory J. Tripoli, HarryWeber, Chun-ChiehWu 3.2.1 Introduction This Topic includes developments related to uni- and multi-model ensemble forecasting, data assimilation, and model developments as they relate to forecasts of tropical cyclone (TC) track. The TC track forecast is a critical component of the warning system, as it serves as the basis for forecasting the areas threatened by damaging winds, storm surge, and rainfall. The track errors are only averages from which there can be large deviations in any individual case, even at short ranges. 3.2.2 The current state of operational TC track forecasts The current state of operational TC track forecasts globally is covered in Topic 3.1. The following is a short discussion of enhancements to operational track forecasts during the last four years. Comparisons between different basins should not be made because of varying numbers of cases and difficulty in forecasting. 3.2.2.1 North Atlantic Ocean Official TC track forecasts in the North Atlantic Ocean were extended from 3 days to 5 days by the United States (US) National Hurricane Center (NHC) in 2003. Table 3.2.1 shows average official NHC track forecast errors from 2001-2005. Five-day forecasts are currently as accurate as the 3-day forecasts 15 years ago, and forecast errors from 24-72 h are now roughly half their 1990 values. The extended forecasts have not been produced long enough to assess any trends in their errors (Fig. 3.2.1). North Atlantic Ocean 0 h 12 h 24 h 36 h 48 h 72 h 96 h 120 h

Track Error (km) 13.7 69.1 119.4 169.1 219.1 317.4 427.9 561.6

Table 3.2.1 Average official NHC track forecast errors in the North Atlantic Ocean from 2001-2005 for all TCs.

391

3.2.2.2 Eastern North Pacific Ocean Official TC track forecasts in the Eastern North Pacific Ocean were extended from 3 days to 5 days by NHC in 2003. Average official NHC track forecast errors from 2001-2005 are shown in Table 3.2.2. The 4-day forecasts are currently as accurate as the 3-day forecasts were 15 years ago (Fig. 3.2.2). These extended forecasts have not been produced for a long enough period of time to assess any trends in their errors. 3.2.2.3. Central North Pacific Ocean Official TC track forecasts in the Central North Pacific Ocean were extended from 3 days out to 5 days by the US Central Pacific Hurricane Center (CPHC) in 2003. Average official CPHC track forecast errors from 2001-2005 are shown in Table 3.2.3. Because of the relative inactivity of the basin, some seasons do not have forecasts reaching 72 h or beyond. These extended forecasts have not been produced for a long enough period of time to assess any trends in their errors (Fig. 3.2.3). Eastern North Pacific Ocean 0 h 12 h 24 h 36 h 48 h 72 h 96 h 120 h

Track Error (km) 17.2 65.0 111.3 152.8 190.0 267.8 355.1 427.9

Table 3.2.2. Average official NHC track forecast errors in the Eastern North Pacific Ocean from 2001-2005 for all TCs.

Figure 3.2.1. Annual average NHC official track errors for North Atlantic Ocean tropical storms and

hurricanes for the period 1970-2005, with least-squares trend lines superimposed.

392

3.2.2.4 Joint Typhoon Warning Center The U. S. Joint TyphoonWarning Center (JTWC) extended their TC track forecasts in the Western North Pacific Ocean, the Southern Pacific Ocean, and the Indian Ocean from 3 days out to 5 days in 2001. Average JTWC track forecast errors from 2001-2005 are shown in Fig. 3.2.4. Though 48-h forecasts are now as accurate as 24-h forecasts were in 1976, and 72-h forecasts are now as accurate as 48-h forecasts were in 1999, these extended forecasts have not been produced for a long enough period of time to assess any trends in their errors. Central North Pacific Ocean 12 h 24 h 36 h 48 h 72 h 96 h 120 h

Track Error (km) 50.4 139.2 187.4 241.4 337.1 376.5 376.2

Table 3.2.3. Average official CPHC track forecast errors in the Central North Pacific Ocean from 2001-2005 for all TCs.

Figure 3.2.2. Annual average NHC official track errors for Eastern North Pacific Ocean tropical storms

and hurricanes for the period 1970-2005, with least-squares trend lines superimposed.

393

3.2.3 Numerical Weather Prediction globally The best objective track guidance tools for TC forecasters are now dynamical models. These models have improved steadily in recent years due to improvements to data, data assimilation, resolution, and vortex initialization. A number of global and limited area (regional) numerical weather prediction models are now available at the various TC centers. Typically, the forecaster must resolve significant differences among the forecasts from these models under limited time constraints. A consensus forecast based on an ensemble of model runs yields better forecasts, on average, than those from any single member of the ensemble. For hurricane track forecasts, the most successful ensembles so far are those made up of independent models, rather than ensembles drawn from multiple runs of a single model. Furthermore, a sense of the reliability of the individual track forecasts is nearly as important as the actual forecasts. Various techniques to improve track forecasts and to provide estimates of their reliability have been devised based upon combining ensembles of tracks from individual and multiple modeling systems. The following sections provide information on these ensembles, followed by information on research and plans for next-generation TC forecasting systems.

Figure 3.2.3. Annual average CPHC official track errors for Central North Pacific Ocean

tropical storms and hurricanes for the period 1997-2005, with least-squares trend lines

superimposed.

394

3.2.3.1 Single-model ensembles Ensemble forecasting techniques are increasingly used in TC track prediction. Single-model ensembles generally perturb initial conditions of the deterministic run that represents the best estimate of the initial state. These perturbations are done either randomly (Cheung and Chan 1999a,b) or in a way that represents analysis uncertainty or maximizes perturbation growth. These latter techniques include singular vectors (Leutbecher 2005), bred-modes (Cheung and Chan 1999a,b, Toth and Kalnay 1997), or an ensemble transform (Wang and Bishop 2003). Benefits of these techniques include state-dependent estimates of perturbation structures that are dynamically constrained and fast growing. Disadvantages include the fact that the dynamically generated perturbations are only approximations to actual analysis errors and the individual single-model ensemble runs are not fully independent of the other runs of the ensemble; this limits the ability of the ensemble to span the space of possible solutions. In recent years, services producing ensemble forecasts have been increasing the number of ensemble members and upgrading the models. Improvements to singular vector techniques and the introduction of Ensemble Kalman Filter (EnKF) techniques to produce initial conditions has occurred. Specifically, ensembles have been used to predict TC tracks, especially to provide probabilistic track forecasts that have been useful to operational forecasters in the medium range. Also, a number of centers are introducing short-range ensembles. Australian Bureau of Meteorology (BoM) runs ensembles of the Limited Area Prediction System (LAPS) incorporating prediction of TC tracks. An ensemble of Weber's Barotropic Model (WBAR) based on perturbations to the steering layer is being developed for use at BoM. Statistical post-analyses of the past season are included in the generation of its deterministic and probabilistic output. In combination with the computational efficiency of a barotropic model in comparison with 3-D models, large statistical

Figure 3.2.4. Annual average JTWC track errors for Western North Pacific Ocean, South Pacific

Ocean, and Indian Ocean tropical cyclones for the period 2001-2005, in nm.

395

samples can be produced that provide statistically-significant estimates of its future deterministic and probabilistic performance quality. Furthermore, the possibility of large numbers of runs allows the model to be used as a benchmark for the general capabilities of single-model ensembles, a task that can hardly be achieved for the more complex models. Analog ensemble forecast model has been developed for tracks in the Australia region (Fraedrich et al. 2003). The Canadian Meteorological Centre (CMC) produces 8 ensemble members twice daily at T149L28 resolution with the global spectral model, and an additional 8 members twice daily at 1.2 degree resolution with GEM, both through 16 days. Because two different models are used, this can also be considered to be a multi-model ensemble. The 16 models have different physical parametrizations, data assimilation cycles and sets of perturbed observations that are assimilated with an Ensemble Kalman Filter with 96 members to define the model error covariance terms. Boundary conditions such as sea surface temperature, albedo and roughness length have been perturbed as well. The control forecast is initiated from the ensemble mean and performed with the spectral model. The ensemble may be soon upgraded to run twice daily. The China Meteorological Administration (CMA) has developed an ensemble forecasting system for TC tracks that will run experimentally during 2006. It is expected to provide track forecasts and forecasts of the area of high winds and rains produced by TCs. The global model is run at T106L19 resolution through 30 days, with perturbations calculated using singular vectors. The European Centre for Medium-RangeWeather Forecasts (ECMWF) singular vector-based perturbations targeted on TCs were revised in 2004. In the old configuration, singular vectors were computed for systems of at least tropical storm intensity between 25S-25N. Furthermore, the optimization regions were centered on the reported TC position and limited to a latitude band from 25S-25N. The perturbations were limited to the tropics in order not to duplicate extratropical singular vectors. However, this resulted in an unrealistic reduction of spread in TC positions during extratropical transitions. Perturbations are now targeted on systems of at least tropical depression intensity between 40S and 40N. A duplication of extratropical perturbations is avoided by computing the singular vectors targeted on a TC in the subspace orthogonal to the extra-tropical singular vectors. Optimization regions are specific for individual TCs if possible and account for the predicted movement of the TC by using track information from the previous ensemble forecast. The maximum number of TC computations has been increased from 4 to 6. Pre-operational testing of the revised perturbations for TCs indicates an improvement of the 5-day strike probability forecasts produced operationally. The ensemble resolution is T399L62, with perturbations at T42L62, and is now run twice per day. The ensemble prediction system will be upgraded in several steps to a variable resolution system which is coupled to an ocean model --- most likely from day 10 onwards (the exact configuration of the 3rd phase of this system has not yet been defined, but very likely it will be T399L62 to 10 days twice daily, T255L62 to 15 days twice daily, and T255L62 to 32 days once a week). The Japan Meteorological Agency (JMA) instituted a global ensemble prediction system in 2003 for the medium range. It is a 25-member global ensemble run at T106L40 through 9 days, and the perturbations calculated with bred modes. Effective point typhoon strike probability products are created using this ensemble. JMA plans to upgrade the system in 2006 to T319L60 with 50 members, and to produce a typhoon-specific ensemble for track prediction in 2007. This typhoon ensemble will run four times daily through 84 h for probabilistic typhoon forecasts. This probabilistic system will help to define operational reliability circles. The Korea Meteorological Administration (KMA) currently runs a 32-member global ensemble at T106L21 resolution to 10 days using breeding of growing modes for perturbations. They expect to increase the resolution to 30 level. KMA is working toward developing a perturbation scheme that fully reflects initial condition uncertainty in the region of the TC. In 2005, the US National Centers for Environmental Prediction (NCEP) improved their global ensemble

396

forecasting system to allow for vortex relocation in the initial conditions. This eliminated problems with erroneous location of the initial vortex and improved the system's reliability. For the 2006 season, the ensemble forecasting system was upgraded to have 14 members four times daily, with a new perturbation technique based on an ensemble transform. The model is run at T126L28 through 384 h. The first upgrade allowed the ensemble mean to improve upon the high-resolution control track forecast for the first time (Fig. 3.2.5) and for an improved relationship between the spread of the ensemble and the actual error of the ensemble mean. A companion short-range ensemble prediction system is also run with regional models, but no special vortex processing is done. One major problem with these ensemble forecasting systems is dissemination of large amounts of data (over 200 Gbytes of data daily from all ensemble forecast systems globally), and the necessity of gathering these data from geographically diverse centers. To address this, the North American Ensemble Forecast System (NAEFS) is a new weather modeling system run jointly by the Meteorological Service of Canada (MSC) and NCEP to provide numerical weather prediction products to weather forecasters in both countries for a forecast period that runs to 2 weeks. The NAEFS combines the Canadian global forecast model ensemble and the NCEP global ensemble into a joint ensemble that will create weather forecasts for all of North America. At present, the national weather agencies in North America are participating in NAEFS - MSC, the National Meteorological Service of Mexico, and NCEP. The THORPEX Asian Regional Committee is arranging the exchange of TC track forecasts among its members. The plan is to develop a multi-model ensemble and deliver probabilistic forecasts to its members by 2008 as part of a proposed THORPEX Interactive Global Grand Ensemble (TC-TIGGE) to accelerate the improvements in the accuracy of 1-day to 2-week high-impact weather forecasts for the benefit of humanity. The goals of TIGGE are: (1) enhanced international collaboration between operational centers and universities on the development of ensemble prediction, (2) development of new methods combining ensembles of predictions from different sources and of correcting for systematic errors (biases, spread over-/under-estimation), (3) increased understanding of the contribution of observation, initial and model uncertainties to forecast error, (4) increased understanding of the feasibility of operationally employing, an interactive ensemble system that responds dynamically to changing uncertainty (including the use of adaptive observing, variable ensemble size, on-demand regional ensembles) and which exploits new technology for grid computing and high-speed data transfer, (5) development of a prototype future Global Interactive Forecasting System, and (6) communication of probabilistic forecasts to the public.

397

3.2.3.2 Multiple-model ensembles The simplest form of ensemble forecast involves post-processing of existing deterministic forecasts produced routinely from operational centers. This technique has the advantage of very low computational overhead (only post-processing) and is fairly successful (Goerss 2000, 2006,Williford et al. 2003, Vijaya Kumar et al. 2003), but suffers from small ensembles (limited to the number of operational centers) and the fact that the ensemble members are not equally likely, independent, realizations. Bias removal techniques are difficult with single-model ensembles since the individual forecasts are not independent, but have proven useful (Williford et al. 2003 and Vijaya Kumar et al. 2003), and Bayesian model averaging (Raftery et al. 2005) would also probably significantly improve the skill of these forecasts. Verifications show that the multiple model ensemble approach results in noticeable improvements in TC forecast track errors (Goerss 2000; Lee andWong 2002). CMC produces 8 ensemble members twice daily at T149L28 resolution with the global spectral model, and an additional 8 members twice daily at 1.2 degree resolution with GEM, both through 16 days. Because two different models are used, this can also be considered to be a multi-model ensemble. See Section 3.2.3.1 for more information. CMA has developed a multi-model ensemble forecasting technique based on canonical correlation analysis. The system utilizes six track forecasts from numerical weather prediction models and from forecasting centers, and calculates mean forecasts based on multivariate linear regression, weighting based on past forecasts, and a simple arithmetic mean. Each of these tracks improve upon any of the individual forecasts. The Hong Kong Observatory is experimenting with ensembles of their mesoscale model system. JMA has used an arithmetic mean of the JMA, United Kingdom Meteorological Office (UKMO) global

Figure 3.2.5. Track forecast error (nm) during 2003 and 2005 for the NCEP Global Forecasting

System (AVNI) and the mean of the Global Ensemble Forecasting System (AEMI).

398

model, and ECMWF track forecasts since 1991. This ensemble mean has performed better than any of the individual components each year since 1996. The NCEP uses multiple techniques for multi-model ensembles. One such technique, GUNA, is the mean of the Geophysical Fluid Dynamics Laboratory (GFDL) Hurricane Model, UKMO, US Navy Operational Global Atmospheric Prediction System (NOGAPS), and US NCEP Global Forecasting System (GFS) forecasts available only when all four models provide tracks. The second technique, CONU, uses tracks from the GFDL, GFS, NOGAPS, GFDN (the GFDL model run by the U. S. Navy using NOGAPS for boundary conditions), and the UKMO, and is available when every individual forecast may not be. Another technique available at JTWC, CONW, includes up to nine model forecasts. Goerss (2006) developed a product to predict the reliability of the ensemble mean forecast based on a stepwise linear regression using predictors from previous seasons. The result is a set of radii at each forecast time for each individual forecast within which the storm center will reside with 75% likelihood. The size of the circles is representative of the likely reliability of the forecast. Results have generally been positive, but this system requires that the verification lie within the spread of the individual models. NHC forecasters have come to rely on these techniques as some of their primary guidance. The National Center for Hydro-Meteorological Forecasting (NCHMF) or Vietnam has used an arithmetic mean of advisory forecasts available from JMA, JTWC, and CMA. This mean forecast has outperformed available model and forecasting center tracks in 2004 and 2005. Sophisticated multi-model ensemble approaches include the Florida State University (FSU) Superensemble and the JTWC Probabilistic Ensemble System for the Prediction of TCs (PEST; Weber 2005). In the Superensemble, past numerical and official TC forecasts are regressed to the observed tracks and intensities for past storms for each forecast time at 12-h intervals through 5 days during a training period. A simple multiple linear regression technique generates weights for each model, and bias estimates are calculated. The coefficients are then used during the forecast phase. Future forecasts made by the multiple models and the aforementioned statistics are used to construct the superensemble forecasts. These forecasts have outperformed other multi-model ensembles in the Atlantic basin during 2004 and 2005 (Fig. 3.2.6). PEST uses all available (early) model forecasts during a training period (the previous season) and produces a deterministic and probabilistic position and intensity prediction by application of the results of a statistical analysis of all model forecasts. The track guidance quality is approximately the same as that of CONW (within ±5% in skill) and is better than all other models in use in the western North Pacific. 3.2.4 Data Assimilation Most operational systems currently use a three-dimensional variational (3D-VAR) or a four -dimensional variational (4D-VAR) assimilation system each analysis cycle. Increased resolution of numerical models and of high-density remotely sensed data make the problem of assimilation increasingly important. For example, an advanced data assimilation system in H-WRF will make use of real time airborne Doppler radar data to initialize the three dimensional storm scale structure. The initial plan is to use a 3D-VAR Gridpoint Statistical Interpolation (GSI) technique to assimilate these data. During the 2006 NOAA Hurricane Field Program, airborne Doppler radar data will be quality controlled on the aircraft and transmitted in real time to NCEP for potential assimilation into the model. Initial attempts at this assimilation have been promising (Liu et al. 2006), and further studies with a 4D-VAR or ensemble-based systems are pending. Preliminary EnKF studies for TCs (Chen and Snyder 2006, Hakim and Torn 2006, Torn and Hakim 2006, Aberson and Etherton 2006) suggest that this technique has significant potential that warrants further research. Hakim and Torn (2006) and Torn and Hakim (2006) explored the use of an EnKFs at 10 km and 30 km horizontal resolution for ensemble analysis and prediction of Hurricanes Katrina (2005) and Rita (2005). All conventional observations were assimilated except satellite radiances.

399

Figure 3.2.6. Track forecast skill for various dynamical models and multi-model ensembles in the Atlantic basin during 2004 and 2005.

400

Preliminary results suggest that the EnKF is able to accurately constrain both analysis and forecast errors. Future research will extend this work to smaller scales and to a larger sample of cases. Aberson and Etherton used an Ensemble Transform Kalman Filter to assimilate environmental dropwindsonde data into a simple barotropic model, changing the initial condition in a way constrained by the model and the ensemble, and improving the ultimate track forecast over what was possible with a 3D-VAR data assimilation system (Fig. 3.2.7). The emergence of ensemble data assimilation methods such as the EnKF brings with it the possibility of casting the more general problem of numerical weather prediction, and specifically involving TCs, in a probabilistic framework. Since any single model forecast is only one of infinitely many possible realizations of a random process, the evolution of which depends strongly on the realizations of other random variables (e.g. the observations which impact the analysis used to initialize the forecast), it is important to focus not only on the forecast itself but also on the associated forecast error (Lewis and Tripoli 2006). Figure 3.2.8 shows the results from a single assimilation cycle and represents the ensemble mean forecast for 1 km wind 18Z on 25 October for Hurricane Wilma. The resulting improvement in the analysis is evident when comparing a) and b) to the verifying field in c). Here, the assimilation process is not viewed as one of obtaining a deterministic analysis, since not even satellite data can ever be expected to resolve basic cumulus cells in space or time. Instead, the assimilation is aimed at a cloud resolving probabilistic ensemble analysis and ultimately a cloud resolving probabilistic ensemble forecast . With these analyses and forecasts, the width of the ensemble determines the probability of the analysis (or forecast) and the mean of the ensemble determines the most likely analysis (or forecast). This eliminates the need to generate an initial ensemble from a deterministic analysis and sets more realistic goals for a cloud resolving analysis of a convective field. Moreover, as a developing TC becomes more focused on the storm scale circulation and is less dependent on on dividual clouds or embedded mesoscale convective systems, the analyses and forecasts will naturally narrow and convey its increasing predictability.

401

Figure 3.2.7 Streamline analysis of the increments to the VICBAR DLM wind field first guess using (a) isotropic and (b) ensemble-based error statistics at 0000 UTC 24 Sep 2001 on dropwindsonde data from a synoptic surveillance mission around Hurricane Humberto (From Aberson and Etherton 2006).

402

3.2.5 Model developments The following is a list of current global and regional models used around the world for TC track forecasting, and recent and planned upgrades. Like with ensemble forecasts, the amount of data produced by the modeling systems is very large. Unfortunately, the different centers do not necessarily share model output with each other so model intercomparisons are difficult. Such comparisons would allow for divining what data assimilation and modeling techniques work best, especially with regard to vortex initialization when data are limited. 3.2.5.1 Global Models The BoM Global Analysis and Prediction (GASP) Model is run at T239L29 resolution twice daily to 7 days. The Global/Regional Assimilation Prediction System (GRAPES) is the CMA next-generation unified numerical weather prediction system that includes a data assimilation system, a regional mesoscale numerical weather prediction model, and a global medium-range numerical weather prediction model.

Figure 3.2.8. Left: 1-km windspeed (kt) valid at 18Z 25 October: a) ensemble mean before assimilation, b) ensemble mean after assimilation, c) verifying analysis. Right: 1-km simulated reflectivity factor (dBZ) valid at 18Z 25 October: a) ensemble mean before assimilation, b) ensemble mean after assimilation, c) verifying analysis.

403

It has a unified multi-scale dynamic core for both the regional and the global models and has physics applicable to both systems. Data are assimilated currently with 3D-VAR, with ongoing development of 4D-VAR and and EnKF. Specifically for typhoon track forecasting, CMA utilizes its Global Model for Typhoon Track Prediction (GMTTP) coupled to the global mediumrange spectral model run at T213L31 resolution. This system became operational model in 2006. It provides 5-day forecasts four times daily for up to three TCs at each time. Typhoons are initialized with a synthetic vortex, but work on a relocation technique is ongoing. The global model itself is also run at T213L31 resolution, and is run through 10 days. The resolution of the ECMWF data assimilation and modeling system was recently increased to T799L90. No special processing near the vortex is used. The India National Center for Medium-RangeWeather Forecasts (NCMRWF) currently runs its global model at T170L28 resolution to five days. The JMA Global Spectral Model (GSM) has a 3D-VAR data assimilation system and is run at a resolution of T213L41 through 216 h. The cyclone is initialized by using an axisymmetric vortex and adding an asymmetry derived from the first-guess field, a 6-h forecast. The model is hydrostatic and is coupled to the sea using a 1-degree resolution analysis with a climatic trend. It is run twice daily (to 90 h at 0000 UTC, and to 216 h at 1200 UTC). TC track forecasts are only made through 90 h. The data assimilation was expected to be upgraded to 4D-VAR in 2005. KMA recently upgraded the Global Data Assimilation and Prediction System (GDAPS) to run to 10 days with a resolution of T426L40 concurrently with an upgrade to the model physics for best performance. This led to an improvement in the representation and forecast of typhoons in the model. MeteoFrance uses its global model ARPEGE, a variable-mesh system with a horizontal resolution as low as 4.23 km and 41 levels in the vertical. Data are assimilated with a 4D-VAR system, and the model is run to 102 h. A special tropical version of ARPEGE for the tropics is run at T384L41 resolution over the Indian Ocean, and is run through 72 h. The MSC Global Environmental Mesoscale (GEM) Model is run on a 0.9-degree grid with 28 levels through 360 h once each day and to 240 h another time each day. Data are assimilated into the model using an EnKF system. The model has proven effective in predicting TC tracks. UKMO runs a non-hydrostatic global model four times daily on a grid with a resolution of 0.556 degrees latitude and 0.833 degrees longitude with 38 levels. Data are assimilated using 4D-VAR. The US GFS is run at T382L64 four times daily through 180 h, and at T190L64 through 2384 h. Data are assimilated with a 3D-VAR system, and the TC vortex is relocated from its position in the first-guess field to its analyzed location at each cycle. The model is part of the primary guidance used at NHC. Future upgrades include a change in the assimilation system from spectral statistical interpolation to a gridpoint statistical interpolation. Recent upgrades to NOGAPS have included an increase in resolution to T239L30, the use of a 3D-VAR technique for data assimilation, and improvements to the synthetic vortex technique. The model is run four time daily providing 5-day TC forecasts globally. 3.2.5.2 Regional Models The BoM TC-LAPS forecast systems provides very reliable forecasts in the Australia and western South Pacific regions. The current model has an inner mesh of 0.15 degree resolution centered on the TC with 29 levels in the vertical. Recent upgrades have involved improved specification of the TC vortex, a new bulk explicit microphysics scheme, and inclusion of data from new sources such as

404

surface winds and heating rates derived from satellite-based instruments. These upgrades are leading toward an ultimate goal of a fully coupled ocean-atmosphere version of the model. In addition to CMA models discussed in Section 3.2.5.1, some provincial centers continue to use statistical-dynamical techniques and mesoscale models such as MM5 to forecast for their area. A triply-nested version of MM5 with a nudging technique for data assimilation and maximum horizontal resolution of 3 km and 36 levels is run through 48 h. The Shanghai Typhoon Institute currently runs GRAPES as a regional model with a resolution of 0.25 degrees and 31 vertical levels. The cyclone is initialized with a 4DVAR bogus data assimilation in which all model fields are adjusted to fit the bogus fields with the constraint of the forecast model. The observations are then assimilated using 4D-VAR. Future work on the model will be to develop new boundary layer and convective parameterization schemes, to couple the atmospheric model with an ocean and a land model, and to improve the 3D- and 4D-VAR schemes, especially in regard to initializing the vortex. The Hong Kong Observatory Operational Regional Spectral Model (ORSM), a doubly-nested mesoscale model with highest resolution of 20 km in the horizontal and 36 levels in the vertical, is run eight times each day to 24 h; the low-resolution (60 km grid) outer mesh is run four times per day to 48 h. Data are assimilated with a threedimensional optimal interpolation scheme, and hourly rainfall data are initialized physically. JMA's GSM is used as boundary conditions. The Observatory is currently experimenting with higher resolution (up to 5 km) non-hydrostatic models with threedimensional or four-dimensional variational and Extended Kalman Filtering data assimilation systems to assimilate high-resolution Doppler data. NCMRWF currently runs a triply-nested version of the MM5 with maximum horizontal resolution of 10 km and 23 levels in the vertical through 5 days, and are planning an upgrade to 42 vertical levels. The NCMRWF global model is used for initial and boundary conditions. NCMRWF also runs a version of the mesoscale Eta model with 48 km resolution and 38 vertical levels through 5 days. Again, the NCMRWF global model is used for initial and boundary conditions. Model output is available to the India Meteorological Department (IMD) for forecasting. For cyclones, a quasi-Lagrangian model with a 3D-VAR data assimilation system is run at 40 km resolution with 16 levels to 36 h. The JMA Typhoon Model (TYM) is run four times daily through 84 h for a maximum of two TCs at each forecast time. The model has a 3D-VAR data assimilation system using a 6-h GSM forecast as its first guess. The cyclone initialization involves an axisymmetric vortex with an asymmetry derived from the previous TYM model run tuned so that the initial motion fits the analyzed track. The vortex is blended with the initial fields using a linear weighting function in an annulus with a maximum size of 800 km thus preserving asymmetric components. The model is hydrostatic and is run at a resolution of ~ 24 km in the horizontal, with 25 levels in the vertical. A 1-degree analysis of sea surface temperature is used for the lower boundary. The KMA Regional Data Assimilation and Prediction System (RDAPS) was recently upgraded to have a larger domain to allow for longer forecasts of systems moving westward. The model is run with a horizontal resolution of 5 km to 24 h and 30 km to 48 h, both with 43 vertical levels, using GDAPS as boundary conditions. Additionally, KMA also runs a typhoon model (KTM) with 1/6-degree horizontal resolution and 18 levels through 72 h, also using GDAPS as boundary conditions. KMA runs a Double-Fourier Series Barotropic Typhoon Model (DBAR). Tests with a nextgenerational regional model based on a version ofWRF with 10 km horizontal resolution and 31 vertical levels is ongoing. The Macao Meteorological and Geophysical Bureau (SMB) Mesoscale Model is a nested version of the MM5 with highest resolution of 18 km in the vertical and 22 horizontal levels to 60 h. The Mexican Meteorological Service (MSM) runs a non-hydrostatic version of the MM5 using the NCEP GFS as boundary conditions. The horizontal resolution is 45 km, and the model is run with 20 levels through 72 h.

405

MeteoFrance uses its fine-mesh limited area model ALADIN over the Indian Ocean in a configuration known asARPEGE-Tropique. The model is run at 31 km resolution over the Indian Ocean using ARPEGE for boundary conditions. The TC is bogussed and the model physics are tuned to the tropics. MSC GEM is run in regional mode on a 15 km grid with 28 levels over North America through 48 h. NHCMF runs two barotropic models (WBAR and BARO) when a TC is in their area of interest. Additionally, NCHMF runs a mesoscale model to three days with a horizontal resolution of 28 km, and 20 levels in the vertical. The Philippines Atmospheric, Geophysical & Astronomical Services Administration (PAGASA) runs the MM5 at 26 km resolution with 36 levels through 72 h using the NCEP GFS as boundary conditions. The US GFDL coupled hurricane model has provided the most reliable track guidance in the Atlantic and Eastern North Pacific Oceans during the 2003 – 2005 seasons. The current model is hydrostatic with a resolution has 42 levels and is 1/12 degrees in the inner mesh. The 2006 season will see a planned upgrade to Ferrier microphysics (a simplified scheme that includes only cloud and rain water and ice and advects only the combined condensate) that is likely to improve forecasts of sheared storms. The addition of dissipative heating due to friction is likely to improve forecasts of strong TCs as its effect increases with the cube of the wind speed. In tests with cases during the 2004 and 2005 seasons, the new model provided track forecasts up to 12% better than the previous version. The U.S. Navy also runs this model using NOGAPS fields as boundary conditions (GFDN). This is likely to be the last major upgrade of the GFDL model before the next-generation HurricaneWeather Research and Forecast (HWRF) model, a non-hydrostatic, high-resolution, coupled air-sea-land prediction system with advanced physics, becomes operational at the NCEP in coming years. 3.2.6 References Aberson, S. D., and B. J. Etherton, 2006: Targeting and data assimilation studies during Hurricane Humberto (2001). J. Atmos. Sci., 63, 175-186. Chen, Y., and C. Snyder, 2005: Assimilating vortex position with an ensemble Kalman filter. Proc., 11th Conf. On Mesoscale Processes, Albuquerque, NM. Cheung, K. W., and J. C. L. Chan, 1999a: Ensemble forecasting of tropical cyclone motion using a barotropic model. Part I: Perturbations of the environment. Mon. Wea. Rev., 127, 1229–1243. Cheung, K. W., and J. C. L. Chan, 1999b: Ensemble forecasting of tropical cyclone motion using a barotropic model. Part II: Perturbations of the vortex. Mon. Wea. Rev., 127, 2617–2640. Evensen, G., 2003: The ensemble Kalman Filter: Theoretical formulation and practical implementation. Ocean Dynamics, 53, 343-367. Fraedrich, K., C. C. Raible, and F. Sielmann, 2003: Analog ensemble forecasts of tropical cyclone tracks in the Australian region. Wea. Forecasting, 18, 3–11. Goerss, J. S., 2000: Tropical cyclone track forecasts using an ensemble of dynamical models. Mon. Wea. Rev., 128, 1187-1193. Goerss, J. S., 2006: Prediction of tropical cyclone track forecast error for Hurricanes Katrina, Rita, and Wilma. Proc., 27th Conf. On Hurricanes and Tropical Meteorology, Monterey, CA.

406

Hakim, G. J., and R. D. Torn, 2006: Tropical cyclone dynamics deduced from ensemble state estimation. Proc., 27th Conf. On Hurricanes and Tropical Meteorology, Monterey, CA. Lee, T. C., and M. S. Wong, 2002: The use of multi-model ensemble technique for tropical cyclone track forecast at the Hong Kong Observatory. WMO Commission for Basic Systems Technical Conference on Data Processing and Forecast Systems, Cairns, Australia. Leutbecher, M., 2005: On ensemble prediction using singular vectors started from forecasts. Mon. Wea. Rev., 133, 3038–3046. Lewis, W. E., and G. J. Tripoli, 2006: Tropical cyclone modeling in a probabilistic framework. Proc., 27th Conf. On Hurricanes and Tropical Meteorology, Monterey, CA. Liu, Q., N. Surgi, S. Lord, W.-S. Wu, D. Parrish, S. Gopalakrishnan, J. Waldrop, and J. Gamache, 2006: Hurricane initialization in HWRF model. Proc., 27th Conf. On Hurricanes and Tropical Meteorology, Monterey, CA. Raftery, A. E., T. Gneiting, F. Balabdaoui and M. Polakowski, 2005: Using Bayesian model averaging to calibrate forecast ensembles. Mon. Wea. Rev., 133, 1155–1174. Torn, R. D., and G. J. Hakim, 2006: Ensemble analyses and predictions of Hurricane Katrina, Proc., 27th Conf. On Hurricanes and Tropical Meteorology, Monterey, CA. Toth, Z., and E. Kalnay, 1997: Ensemble forecasting at NCEP and the breeding method. Mon. Wea. Rev., 125, 3297–3319. Vijaya Kumar, T. S., T. N. Krishnamurti, M. Fiorino and M. Nagata. 2003: Multimodel superensemble forecasting of tropical cyclones in the Pacific. Mon. Wea. Re., 131, 574–583. Wang, X., and C. H. Bishop, 2003: A comparison of breeding and ensemble transform Kalman filter ensemble forecast schemes. J. Atmos. Sci., 60, 1140–1158. Weber, H. C., 2005: Probabilistic prediction of tropical cyclones. Part I: Position. Mon. Wea. Rev., 133, 1840–1852. Williford, E., C., T. N. Krishnamurti, R. C. Torres, S. Cocke, Z. Christidis and T. S. Vijaya Kumar. 2003: Real-time multimodel superensemble forecasts of Atlantic tropical systems of 1999. Mon. Wea. Rev., 131, 1878–1894.

407

INTENTIONAL LEAVE BLANK

408

INTENTIONAL LEAVE BLANK

409

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES Topic 3.3 : Targeted observation and data assimilation in track prediction Rapporteur: Chun-Chieh Wu PSA Email : cwu@typhoon.as.ntu.edu.tw Fax: 886-2-23632303 Working Group: Sim Aberson, Brian Etherton, Sharan J. Majumdar, Melinda S. Peng, Michael Morgan, Steve Tracton, Zhaoxia Pu, Samuel Westrelin, and Munehiko Yamaguchi 3.3.1. Introduction The objective of this report is to document recent progress since IWTC-5 on the topic related to the Targeted observation and data assimilation in track prediction. The report begins by reviewing the background of targeted observations, followed by an introduction to the techniques specifically used for targeted observations and data assimilation to improve tropical cyclone track prediction. These findings are discussed along with some tentative recommendations on these important scientific issues to the workshop. To optimize limited aircraft and satellite resources, making targeted observations only in the critical areas that will have the maximum influence on numerical weather forecasts of tropical cyclones (TC) is important. Therefore, targeted observing strategies for aircraft missions and satellite products must be developed. The prerequisite for devising the observing strategy is to identify the sensitive areas that will have the greatest influence in improving the numerical forecast, or minimizing the forecast error. In aircraft missions in the early 1990s, sensitive areas were subjectively decided through synoptic analysis or some limited numerical model sensitivity tests. For example, the areas around the surface low centers or the upper-level jet streams were considered to be the sensitive areas that needed to be observed to affect the atmospheric characteristics around the weather systems. It has been demonstrated (Burpee et al. 1996 and Aberson and Franklin 1999) that these basic observing strategies resulted in notable improvements in model predictions of tropical cyclones. Rather than using subjective analysis, scientists have recently developed some objective techniques (e.g., adjoint method, singular vectors, and various usages of ensemble forecast system (EFS)) to design targeted observations. These techniques can not only test the atmospheric predictability of numerical models, but also identify the sensitive areas at the model initial time. It is expected that taking extra observations in these sensitive areas can reduce the error of the model forecast. Pu et al. (1998) used the adiabatic version of National Centers for Environmental Prediction (NCEP) operational T126/L28 global spectral model to study the forecast sensitivity to initial analysis differences by using the adjoint method and quasi-inverse linear method. They found that these two methods were somewhat complementary. The quasi-inverse linear sensitivity is reliable in pinpointing the region of origin of a forecast difference, and this is particularly useful for cases in which the ensemble forecast spread indicates a region of large uncertainty, or when a specific region requires improved forecasts. The adjoint sensitivity is useful for identifying areas that have maximum impact on the region of interest, but are not necessarily the regions actually leading to observed differences. Pu et al. propose that both methods can be useful for targeted observing systems.

410

To address the life cycle of cyclones evolving over the North Atlantic Ocean, the Fronts and Atlantic Storm-Track Experiment (FASTEX) was held from January to February 1997 (Joly 1997). The targeted observation strategies included both the subjective analysis and objective sensitive areas for this experiment. The objective sensitivity studies were carried out on the basis of several products (Emanuel and Langland 1998): singular vectors calculated by the adjoint of the linear tangent model of the Navy Operational Global Atmospheric Prediction System (NOGAPS) model (Gelaro et al. 1999); perturbation mean square enstrophy (mean square vorticity) integrated over the 800-900 hPa layer by the tangent linear and adjoint model of the Météo- France Numerical Weather Prediction model (Bergot et al. 1999): singular vectors from the European Center for Medium-range Weather Forecasts (ECMWF) (Montani et al. 1999); and ensemble transform method from the NCEP global model (Bishop and Toth 1999). An interagency field program called the North Pacific Experiment (NORPEX) during the winter of 1998 directly addressed the issue of observational sparsity over the North Pacific basin, which is a major contributing factor for short-range forecast failures of landfalling Pacific winter systems that affect the United States (Langland et al. 1999). The objective targeting in NORPEX was built upon experiences gained during the FASTEX. However, the targeted observations were performed using two objective methods: singular vectors computed using the NOGAPS and the ensemble transformation applied to the NCEP and ECMWF global ensemble forecasts. 3.3.2 Recent techniques for targeted observations for tropical cyclones For operational surveillance missions in Atlantic hurricanes conducted by the NOAA (Aberson 2003) and the DOTSTAR (Dropwindsonde Observations for Typhoon Surveillance near the Taiwan Region; Wu et al. 2005), four sensitivity techniques have been used to determine the observation strategies. 3.3.2.1 Deep-Layer Mean wind variance Based on the deep-layer mean (DLM) or 850-200 hPa steering flows from the NCEP EFS Global Ensemble Forecasting System (EFS; Aberson 2003), areas with the largest (DLM) wind ensemble spread represent the sensitive regions at the observing time. The DLM wind ensemble spread is chosen because tropical cyclones are generally steered by the environmental DLM flow, and the dropwindsondes from the NOAA Gulfstream IV sample this flow. Aberson (2003) demonstrated that using only the subset of observations in the areas of high NCEP EFS DLM wind variance improved the TC track forecasts more than using uniformly-sampled observations. While the variance in the DLM wind may amplify in the model and be propagated into a chosen verification region at some future time, it may also decay. Aberson (2003) showed that this strategy selected structures at the observing time that most significantly reduced the forecast error variance near the tropical cyclone, but not necessarily within a particular verification region at a verifying time. An example for this DLM method is shown in Fig. 3.3.1 for Typhoon Meari (2004) in the western North Pacific. The DLM wind variance indicates that, at the planned observing time of 1200 UTC 25 Sep 2004, the sensitive areas will be to the northeast and southwest of the center of Meari.

411

Fig. 3.3.1 NCEP Deep-Layer Mean (DLM) wind variance for Typhoon Meari (2004) at 1200 UTC 25 Sep 2004. The black dots represent the locations where the dropwindsondes were deployed in a DOTSTAR mission.

412

3.3.2.2 Ensemble Transform Kalman Filter (ETKF)

This technique predicts the reduction in forecast error variance for feasible deployments of targeted observations based, in this case, on the NCEP EFS (Majumdar et al. 2006). The ensemble transform Kalman Filter (ETKF) (Bishop et al. 2001), like the first technique, uses the differences between ensemble members to estimate regions for observation missions. The ETKF takes the approach of DLM wind variance further. Though DLM wind variance indicates areas of forecast uncertainty at the observation time, it does not correlate initial condition uncertainty with the errors in the forecasts. The ETKF explicitly correlates errors at the observing time with errors of the forecasts. That is, the ETKF identifies ensemble variance that impacts the forecasts in the verifying area at the verifying time. The ETKF uses operational ensemble forecast perturbations to reduce forecast error variance that would be achieved by targeted observations. The analysis error covariance matrix Pr(to)at the observing time (to) pertaining to the routine observational network comprised of rawinsondes and satellite-based temperature fields is found by solving the Kalman filter error statistics equation:

Pr(to)=Pi(to)-Pi(to)HrT(HrPi(to)HrT+Rr)-1HrPi(to), (1)

where Hr and Rr are the observation operator and error covariance matrices, respectively, and Pr is the analysis error covariance matrix at the initial time. The analysis error covariance matrix Pq(to) for the observational network augmented by the qth hypothetical “test-probe” of targeted observations with operator Hq and error covariance matrix Rq is then expressed as

Pq(to)=Pr(to)-Pr(to)HqT(HqPr(to)HqT+Rq)-1HqPr(to). (2)

The associated “signal covariance” matrix valid at the verification time (tv) is given by

Sq(tv)=MPrHqT(HqPrHqT+Rq)-1HqPrM, (3)

where M propagates perturbations from to to tv. The ensemble forecast perturbations at tv are used to rapidly compute the trace of Sq(tv) in the verification region. The ETKF “summary map” represents this signal variance as a function of the central location of adjacent 3×3 test-probes at 1o lat./long. resolution. The test-probe location that produces the highest signal variance is deemed optimal for targeting. For the observing mission of Typhoon Meari (2004), the ETKF technique indicates that in addition to the area around the typhoon center some sensitive areas appear in the northern part of Meari (Fig 3.3.2).

413

Fig. 3.3.2 As in Fig. 3.3.1, except for the NCEP ETKF summary map for Typhoon Meari (2004). 3.3.2.3 Singular Vector (SV) technique The SV technique maximizes the growth of a total energy or kinetic energy norm (e.g., Palmer et al. 1998; Peng and Reynolds 2006) using the adjoint and forward-tangent models of the Navy Operational Global Atmospheric Prediction System (NOGAPS; Rosmond 1997; Gelaro et al. 2002), and also the ensemble prediction system (EPS) of Japan Meteorological Agency (JMA). The leading singular vector (SV) represents the fastest growing perturbation to a given trajectory (such as a weather forecast) in a linear sense (Peng and Reynolds 2006). Consider a nonlinear model M, acting on a state vector x, such that M(x0)=xt, where the subscript refers to the integration time. Let x0

’ represent some perturbed initial state, such that x0

’- x0=p0 and M(x0’)-M(x0)=pt. For linear

perturbation growth, the initial perturbation can be propagated forward in time using the forward-tangent propagator, L, representing the model equations of M linearized about the nonlinear trajectory, such that

≅0 tLp p

. (4)

414

Then L can be represented by its singular values and initial- and final-time SVs as :

1/ 2 / 2−= T 1L E UDV E , (5)

where V (U) are matrices with columns composed of the initial (final) SVs, and D is a diagonal matrix whose elements are the singular values of L. Here E is the matrix that defines how the perturbations are measured. Thus, the SVs form an E-orthonormal set of vectors at the initial and final times. The SVs satisfy the eigenvector equation LTELyn=dn

2Eyn where yn=E-1/2vn, and dn and vn are the nth singular value and initial-time SV, respectively. The leading SV maximizes the ratio of the final perturbation energy to the initial perturbation energy:

t t

0 0

p ;Epp ;Ep (6)

where < > represents a Euclidean inner product. The 2nd SV maximizes this ratio under the constraint of being orthogonal to the first SV, the 3rd SV maximizes this ratio under the constraint of being orthogonal to the first two SVs, and so on. For complex models such as dynamical tropical cyclone models, the eigenvector equation may be solved in an iterative fashion using the forward and adjoint propagators linearized about a particular forecast. The sensitivity areas for Typhoon Meari (2004) determined from the NOGAPS SV method are shown in Fig. 3.3.3. The sensitive areas at the observing time are in the northern and southern regions of Meari, with maximum values to the south of the typhoon center. The sensitive area far to the northwest represents the location of the mid-latitude trough at that time. It is expected that this mid-latitude trough may have some influence on the future track of Meari. Another singular vector method is calculated from the JMA EPS (Yamaguchi, personal communication 2006). A typhoon EPS that is planned to be operational in 2007 has been developed using the linearized model and its adjoint version adopted for the JMA global 4D-Variational analysis system. The linearized model and adjoint consist of full dynamics based on Eulerian integrations and full physical processes including representations of vertical diffusion, gravity-wave drag, large-scale condensation, long-wave radiation, and deep cumulus convection. Two kinds of singular vectors can be calculated: dry and moist singular vectors. Dry singular vectors, which are expected to identify the most unstable dynamical modes of the atmosphere such as a baroclinic mode, are obtained using simplified physical processes that only include vertical diffusion. Moist singular vectors are acquired using full physics, and thus require nearly twice as much computation costs as for the dry ones but capture of the uncertainty in areas such as a tropical region or typhoon surroundings where moist processes are dominant.

415

Fig. 3.3.3 As in Fig. 3.3.1, except for the NOGAPS Singular Vector approach for Typhoon Meari (2004). The sensitive areas for the JMA moist SV method for Typhoon Etau (2003) are given in Figure 3.3.4. The maximum values are in the southwest quadrant of the typhoon. At least with the first 25 singular vectors (Fig. 3.3.5), the JMA dry SV does not have a similar structure around the tropical cyclone.

Fig. 3.3.4 JMA moist Singular Vector for Typhoon Etau (2003) (from Yamguchi).

416

Fig. 3.3.5 JMA (right) dry singular vectors, and (left) moist singular vectors for Typhoon Etau (2003) (from Yamaguchi).

3.3.2.4 Adjoint-Derived Sensitivity Steering Vector (ADSSV) By appropriately defining the response functions to represent the steering flow at the verifying time, a simple vector (ADSSV, defined below)has been designed to clearly demonstrate the sensitivity locations and the critical direction of the typhoon steering flow at the observing time (Wu et al. 2006). Because the goal is to identify the sensitive areas at the observing time that will affect the steering flow of the typhoon at the verifying time, the response function is defined as the DLM wind within the verifying area. A 600 km by 600 km square area centered on the MM5-simulated storm location at the verifying time is used to calculate the background steering flow as defined by Chan and Gray (1982). Two responses functions are then defined: R1, which the 850-300 hPa deep-layer area average (Wu et al. 2003) of the zonal component (u); and R2, the average of the meridional component (v) of the wind vector, i.e.,

∫ ∫∫ ∫≡ hPa

hPa A

hPa

hPa A

dxdydp

dxdydpuR 300

850

300

8501 , and

∫ ∫∫ ∫

≡hPa

hPa A

hPa

hPa A

dxdydp

dxdydpvR

300

850

300

8502 . (4)

By averaging, the axisymmetric component of the strong cyclonic flow around the storm center is removed, and thus the vector of (R1, R2) represents the background steering flow across the storm center at the verifying time. To interpret the physical meaning of the sensitivity, a unique new parameter called the Adjoint-Derived Sensitivity Steering Vector (ADSSV) is designed that relates the sensitive areas at the observing time to the steering flow at the verifying time. The ADSSV with respect to the vorticity field (ς ) is

∂∂

∂∂

≡ςς

21 ,RR

ADSSV , (5)

where the magnitude of ADSSV at a given point indicates the extent of the sensitivity, and the direction of the ADSSV represents the change in the response of the steering flow due to a vorticity perturbation placed at that point. For example, if at a given forecast time the ADSSV vector at one particular grid point points to the east, an increase in the vorticity at this point at the observing time would be associated with an increase in the eastward steering of the storm at the verifying time.

417

The ADSSV based on the MM5 forecast (Fig. 3.3.6) mostly extends about 300-600 km from the north to the east of Typhoon Meari. The directions of the ADSSVs in these areas indicate greater sensitivity in affecting the meridional component of the steering flow.

Fig. 3.3.6 The sequence of ADSSVs based on the -12 h, -24 h, and -36 h MM5 forecasts for Typhoon Meari (2004). 3.3.3. Comparison of targeted observing guidance for track prediction Targeted dropwindsonde observations have been collected in the synoptic environment of hurricanes (Aberson 2003) and typhoons (Wu et al. 2005) to improve operational track forecasts. Although these dropwindsonde observations have been shown to have a positive impact in improving the accuracy of global model track forecasts, the scientific basis for how observations influence forecasts of TC motion and structure remains unexplored. Furthermore, the benefits of assimilating additional observations from different platforms on multiple spatial and temporal scales using novel data assimilation methods have not been adequately studied. Techniques to “target” observations to improve TC forecasts require an accurate representation in the numerical model, data assimilation scheme, and the error propagation and growth. Although several techniques are being tested and compared (Majumdar et al. 2006), they all require evaluation and an improved scientific understanding in the TC environment. Majumdar et al. (2006) conducted a detailed comparison of different targeting techniques. Based on comparisons for 78 two-day forecasts of Atlantic tropical cyclones during the 2004 season, they found that the ECMWF and NOGAPS TESVs (total energy singular vector) offer similar guidance for adaptive sampling on large scales, although smaller-scale aspects local to the tropical cyclone may differ. For major hurricanes, the ETKF and TESV guidance usually both indicate that the optimal locations for adaptive sampling are in a region around the storm. In contrast, the ETKF and TESV guidance is often in considerable disagreement for weaker storms. The sensitivity areas from the ECMWF ETKF resembled those for the NCEP ETKF more for major hurricanes than for weak tropical cyclones, and the NCEP ETKF areas often resembled those based on the NCEP DLM Wind Variance. In a data denial study using the NCEP global model looking at these techniques in the Atlantic, 31 cases from 2004 and 2005 were studied; in each case, the operational cycle with all dropwindsonde data (GSAL) and a parallel run with none of the dropwindsonde data (GSNO) are used. Three additional runs were completed: (i) only those dropwindsonde data that meet the targeting

418

requirements specified in Aberson (2003) are assimilated (GSTG); (ii) those dropwindsondes that meet the sampling strategy specified in Aberson (2003), but with targets defined by the ETKF are assimilated (GSET); and (iii) those dropwindsondes that meet the sampling strategy specified in Aberson (2003), but with targets defined by the NOGAPS singular vectors are assimilated (GSSV). From Table 3.3.1, the 84-h forecasts with ETKF targeting are statistically far better than those from the standard sampling at the 95% level. However, the 12-h and 24-h forecasts with ETKF targeting are statistically far worse than those from the standard sampling at the 90% level. Table 3.3.1 Average track errors (km) for a homogeneous sample of GSET, GSAL, and GSNO.

FCST TIME 12h 24h 36h 48h 60h 72h 84h 96h 108h 120h

GSAL 41.2 69.6 107.7 127.8 182.5 256.8 291.5 348.3 369.9 385.7

GSNO 67.1 93.7 134.1 169.3 193.3 265.5 239.2 299.1 316.8 356.2

GSET 46.2 80.5 114.4 143.0 176.7 239.9 257.3 326.8 356.2 399.4

#CASES 23 23 22 21 20 20 19 18 16 14

For targeting based on the the DLM wind variance, only the 108-h forecasts with ensemble spread targeting (GSTG) are statistically far better than those from the standard sampling (GSAL) at the 95% level (Table 3.3.2). However, the inclusion of all dropsondes (GSAL) and the GSTG targeted approach both lead to significantly poorer track forecasts between 84 h and 120 h. Table 3.3.2 Average track errors of GSTG and comparison with the GSAL and GSNO.

FCST TIME 12h 24h 36h 48h 60h 72h 84h 96h 108h 120h

GSAL 48.4 76.9 110.7 140.6 207.5 289.6 327.0 397.1 459.6 523.0

GSNO 68.1 89.2 123.1 163.5 210.4 292.8 271.4 328.2 339.5 388.1

GSTG 51.5 82.2 111.3 148.2 219.0 289.9 307.3 388.0 413.9 506.1

#CASES 21 21 21 21 21 21 19 17 14 12

In the comparison of the DLM wind variance and the ETKF method together (Table 3.3.3), the ensemble spread targeting is not statistically far different from the ETKF targeting at the 90% level at any forecast time. Table 3.3.3 Average track error (km) for a homogeneous sample of GSET, GSTG, and GSAL and GSNO.

FCST TIME 12h 24h 36h 48h 60h 72h 84h 96h 108h 120h

GSAL 46.6 78.9 114.5 149.3 210.0 300.5 345.6 419.2 456.0 509.9

GSNO 74.1 104.4 138.0 186.6 233.3 313.1 272.4 335.8 362.3 410.8

GSET 53.4 87.4 110.0 149.8 208.4 284.4 306.5 391.5 440.6 526.4

GSTG 52.6 86.7 110.5 151.3 214.1 283.9 311.2 392.6 418.0 499.8

#CASES 15 15 15 15 15 15 14 13 11 9

419

Only two cases with the NOGAPS SVs have been run, not enough to calculate statistical significance. The results of these tests suggest that the targeting and sampling strategy described in Aberson (2003) and from the ETKF are appropriate for the design of flight tracks to improve the tropical cyclone track forecast. Both techniques provide forecasts that are statistically significantly better than those with no sampling during the first three days of the forecast. The sample sizes are larger than those in the HRD synoptic flow experiment (Burpee et al. 1996), and are therefore expected to be robust. Although the singular vector technique is showing promise, it was for a very limited sample of cases. Etherton et al. (2006) qualitatively discussed the observational sensitivity results in 2005 Atlantic season for three strategies: DLM wind variance, ETKF, and ADSSV. The DLM wind variance approach tends to produce sensitivity areas that are very near the center of the tropical cyclone. The ETKF tends to produce a bit more information regarding which features, other than the tropical cyclone, are important to the future track forecast. By construction, the ADSSV rarely, if ever, selects targets in the immediate vicinity of the center of the tropical cyclone. Instead, a ring around the storm is usually the target area, although areas to the south, west, and east are more common than locations to the north of the center of a cyclone. 3.3.4 Further discussions

As shown by Aberson and Etherton (2006) and Huang and Wu (2006), both the targeted observations and the appropriate assimilation of data [such as the advances in the bogused vortex scheme with variational data assimilation (Xiao et al. 2000; Zou and Xiao 2000; Pu and Braun 2001; park and Zou 2004; Wu et al. 2006) and the ensemble transform Kalman filtering method for hurricane study (Aberson et al. 2006 )] from different platforms [such as satellites (GOES by Zou et al. 2001; TRMM by Pu et al. 2004; Bauer et al. 2006a, b) and field programs (CAMEX-4, Kamineni et al. 2006)] with different spatial and temporal resolutions and error characteristics (Fisher 2003; Berre et al 2006, Westrelin et al 2006) will play a very important role in improving tropical cyclone track forecasts. Further development and comparison of the targeting strategies and the improvement of the data assimilation techniques remain important tasks for the future. Note that there has been significant progress in our understanding of the basic oceanic and atmospheric processes that occur during the passage of TCs. Central to these assertions is the need to isolate fundamental physical processes involved in the interactions through detailed process studies using experimental, empirical, theoretical and numerical approaches. As found from measurements, these approaches are needed to improve predictions of not only TC tracks, but also intensity and structure. Some general issues of concerns from the THORPEX report (provided by Steve Tracton) on targeted observations are worth noting here: 1) Although the impact of observation is greater when selected in a sensitive area, the few observations deployed cannot make a substantial impact on the forecasts. 2) The statistical evaluation of the significance of the measured impact requires a large number of cases. 3) Current diagnostics used to evaluate forecasts provides a good assessment of the validity of forecasts (skill), but it may not be sufficient to reveal whether these improvements are relevant to applications (value).

420

4) The use of climatological sensitivities may lead to improvements on average and be more cost effective than targeted observations on demand. 5) Overall, there was a considerable question as to the value of targeting, especially when isolated from the more general issues of observing system sensitivities in design of an “optimal” mix of available observing platforms. 3.3.5 Recommendations Tentative recommendations in this topic group are outlined below, though further elaboration from the workshop is needed: 1) More emphasis on the dependence of targeting impacts upon the data assimilation system and expectations/limitations. This is especially critical in the tropics, where data assimilation is handicapped by weakly balanced flow dependent structure functions and larger (than extratropics) uncertainties/deficiencies in the physics of the assimilating forecast model, especially in regard to high resolution regional systems. 2) More studies of varying definitions, interpretations, and significance of sensitive regions (e.g., different methods, metrics) 3) More work on sampling strategies in sensitive areas, e.g., immediate storm environment for shorter range prediction versus remote areas relevant to longer range forecasts – including the impact of large scales in meso-scales models. 4) More work on metrics to assess the impact of targeting – or more generally on any changes in the observation network. In particular we need to look at the impacts on the variability in ensemble predictions. Thus, for example, track error is only a first order measure – single deterministic forecast which may or may not be representative of an ensemble of forecasts. The end game is whether the observation sensitivity translates to decreased uncertainty (narrower envelope about track – ensemble spread is (should be) more related to the level of uncertainty than skill of any particular single forecast run). 5) Emphasis of the potential value of OSEs and OSSE’s (e.g., Wu et al. 2006) in assessing potential observing system impacts prior to actual field programs. 6) Stronger efforts to develop alternative observing platforms (other than the dropwindsondes) for targeting, especially adaptively selecting satellite observations by revising the data thinning algorithms currently used. Acknowledgements: C.-C. Wu gratefully acknowledges support from the NSC Grant 94-2119-M-002-005-AP1, and ONR-NICOP, Grant N00014-05-1-0672.

421

References Aberson, S. D., 2003: Targeted observations to improve operational tropical cyclone track forecast guidance. Mon. Wea. Rev., 131, 1613-1628. Aberson, S. D., and J. L. Franklin, 1999: Impact on hurricane track and intensity forecast of GPS dropsonde observations from the first-season flights of the NOAA Gulfstream-IV jet aircraft. Bull. Amer. Meteor. Soc., 80, 421-427. Aberson, S. D., and B. Etherton, 2006: Targeting and data assimilation studies during Hurricane Humberto (2006). J. Atmos. Sci., 63, 175-186. Bauer P., P. Lopez, A. Benedetti, D. Salmond and E. Moreau, 2006a : Implementation of 1D+4D-Var Assimilation of Precipitation Affected Microwave Radiances at ECMWF, Part I: 1D-Var. ECMWF Technical Memorandum, 487, 31 pp. Bauer P., P. Lopez, D. Salmond, A. Benedetti, S. Saarinen and M. Bonazzola, 2006b : Implementation of 1D+4D-Var Assimilation of Precipitation Affected Microwave Radiances at ECMWF, Part II: 4D-Var. ECMWF Technical Memorandum, 488, 29 pp. Bergot, T., 1999: Adaptive observations during FASTEX: A systematic survey of upstream flights. Quart. J. Roy. Meteor. Soc., 125, 3271–3298. Berre, L., S.E., Stefanescu, and M. Belo Pereira, 2006: The representation of the analysis effect in three error simulation techniques. Tellus, 58, 196-209. Bishop, C. H., and Z. Toth, 1999: Ensemble transformation and adaptive observations. J. Atmos. Sci., 56, 1748-1765. Bishop, C. H., B. J. Etherton, and S. J. Majumdar, 2001: Adaptive sampling with the ensemble transform Kalman filter. Part I: Theoretical aspects. Mon. Wea. Rev., 129, 420-436. Burpee, R. W., J. L. Franklin, S. J. Lord, R. E. Tuleya, and S. D. Aberson, 1996: The impact of omega dropsondes on operational hurricane track forecast models. Bull. Amer. Meteor. Soc., 77, 925-933. Chan, J. C.-L., and W. M. Gray, 1982: Tropical cyclone movement and surrounding flow relationship. Mon. Wea. Rev., 110, 1354-1376. Emanuel, K., and R. Langland, 1998: FASTEX adaptive observations workshop. Bull. Amer. Meteor. Soc., 79, 1915–1919. Etherton, B., C.-C. Wu, S. J. Majumdar, and S. D. Aberson, 2006: A comparison of the targeting techniques for 2005 Atlantic tropical cyclones. Preprints, 27th Conf. on Hurricanes and Tropical Meteorology, Monterey, CA, Amer. Meteor. Soc. Fisher, M., 2003: Background error covariance modelling. In Recent developments in data assimilation for atmosphere and ocean, ECMWF Seminar proceedings, 45-63. Gelaro, R., R. Langland, G. D. Rohaly, and T. E. Rosmond, 1999: An assessment of the singular vector approach to targeted observing using the FASTEX data set. Quart. J. Roy. Meteor. Soc., 125, 3299-3328. Gelaro, R., T. E. Rosmond, and R. Daley, 2002: Singular vector calculations with an analysis error variance metric. Mon. Wea. Rev., 130, 1166-1186.

422

Huang, W.-P., and C.-C. Wu, 2006: The impact of the dropwindsonde data from DOTSTAR on the track prediction of Typhoon Conson (2004). Preprints, 27th Conf. on Hurricanes and Tropical Meteorology, Monterey, CA, Amer. Meteor. Soc. Joly, A., D. Jorgensen, M. A. Shapiro, A. Thorpe, P. Bessemoulin, K. A. Browning, J.-P. Cammas, J.-P. Chalon, S. A. Clough, K. A. Emanuel, L. Eymard, R. Gall, P. H. Hildebrand, R. H. Langland, Y. Lemaître, P. Lynch, J. A. Moore, P. O. G. Persson, C. Snyder and R. M. Wakimoto, 1997: The Fronts and Atlantic Storm-Track Experiment (FASTEX): Scientific Objectives and Experimental Design. Bull. Amer. Meteor. Soc., 78, 1917–1940. Kamineni R., T. N. Krishnamurti, S. Pattnaik, E. V. Browell, S. Ismail and R. A. Ferrare. 2006: Impact of CAMEX-4 Datasets for Hurricane Forecasts Using a Global Model. J. of Atmos. Sci., 63, 151–174. Langland, R. H., Z. Toth, R. Gelaro, I. Szunyogh, M. A. Shapiro, S. J. Majumdar, R. Morss, G. D. Rohaly, C. Velden, N. Bond, and C. H. Bishop, 1999: The North Pacific Experiment, NORPEX-98: Targeted observations for improved North American weather forecasts. Bull. Amer. Meteor. Soc., 80, 1363-1384. Leidner, S. M., L. Isaksen and R. N. Hoffman, 2003: Impact of NSCAT Winds on Tropical Cyclones in the ECMWF 4DVAR Assimilation System. Mon. Wea. Rev., 131, 3–26. Majumdar, S. J., S. D. Aberson, C. H. Bishop, R. Buizza, M. S. Peng, and C. A. Reynolds, 2006: A comparison of adaptive observing guidance for Atlantic tropical cyclones. Mon. Wea. Rev., in press. Montani, A., J. A. Thorpe, R. Buizza, and P. Unden, 1999: Forecast skill of the ECMWF model using targeted observations during FASTEX. Quart. J. Roy. Meteor. Soc., 125, 3219–3240. Palmer, T. N., R. Gelaro, J. Barkmeijer, and R. Buizza, 1998: Singular vectors, metrics, and adaptive observations. J. Atmos. Sci., 55, 633-653. Park, K. and X. Zou, 2004: Toward Developing an Objective 4DVAR BDA Scheme for Hurricane Initialization Based on TPC Observed Parameters. Mon. Wea. Rev., 132, 2054–2069. Peng, M. S., and Reynolds, C. A., 2006: Sensitivity of tropical cyclone forecasts. Submitted to J. Atmos. Sci. Pu, Z. X., S. J. Lord, and E. Kalnay, 1998: Forecast sensitivity with dropsonde and targeted observations. Tellus, 50A, 391-410.

Pu, Z.-X. and S. A. Braun, 2001: Evaluation of Bogus Vortex Techniques with Four-Dimensional Variational Data Assimilation. Mon. Wea. Rev., 129, 2023–2039.

Pu, Z.-X., W.-K. Tao, S. Braun, J. Simpson, Y. Jia, J. Halverson, W. Olson and A. Hou, 2002: The Impact of TRMM Data on Mesoscale Numerical Simulation of Supertyphoon Paka. Mon. Wea. Rev., 130, 2448–2458. Rosmond, T. E., 1997: A technical description of the NRL adjoint model system, NRL/MR/7532/97/7230, Naval Research Laboratory, Monterey, CA, 93943, 62 pp. Westrelin S., G. Faure, L. Berre, J.-M. Willemet : Test of a mesoscale model over the South West Indian ocean for cyclone analysis and prediction., AMS, 27th Conference on hurricanes and tropical meteorology, Monterey, 24-28 april 2006. Wu, C.-C., T.-S. Huang, W.-P. Huang, and K.-H. Chou, 2003: A new look at the binary interaction:

423

Potential vorticity diagnosis of the unusual southward movement of Typhoon Bopha (2000) and its interaction with Typhoon Saomai (2000). Mon. Wea. Rev., 131, 1289-1300. Wu, C.-C., P.-H. Lin, S. D. Aberson, T.-C. Yeh, W.-P. Huang, J.-S. Hong, G.-C. Lu, K.-C. Hsu, I.-I. Lin, K.-H. Chou, P.-L. Lin, and C.-H. Liu, 2005: Dropwindsonde Observations for Typhoon Surveillance near the Taiwan Region (DOTSTAR): An overview. Bull. Amer. Meteor. Soc., 86, 787-790. Wu, C.-C., J.-H. Chen, P.-H. Lin, and K.-H. Chou, 2006: Targeted observations of tropical cyclone movement based on the adjoint-derived sensitivity steering vector. J. Atmos. Sci. (with minor revision). Wu, C.-C., K.-H. Chou, Y. Wang and Y.-H. Kuo, 2006: Tropical cyclone initialization and prediction based on four-dimensional variational data assimilation. J. of Atmos. Sci., 63, 2383–2395. Xiao, Q., X. Zou and Bin Wang, 2000: Initialization and Simulation of a Landfalling Hurricane Using a Variational Bogus Data Assimilation Scheme. Mon. Wea. Rev., 128, 2252–2269. Zou, X. and Q. Xiao, 2000: Studies on the Initialization and Simulation of a Mature Hurricane Using a Variational Bogus Data Assimilation Scheme. J. Atmos. Sci., 57, 836–860. Zou, X., Q. Xiao, A. E. Lipton and G. D. Modica, 2001: A Numerical Study of the Effect of GOES Sounder Cloud-Cleared Brightness Temperatures on the Prediction of Hurricane Felix. J. Appl. Meteor., 40, 34–55.

424

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES

Special Focus Topic 3a : Sharing experiences in operational consensus forecasting. Rapporteur: Andrew Burton Regional Manager, Severe Weather Services Australian Government Bureau of Meteorology Western Australia Regional Office PO Box 1370, West Perth WA 6872, AUSTRALIA Telephone: +61 8 92632282 Email: A.Burton@bom.gov.au Working Group: Philippe Caroff (RSMC La Reunion - France), James Franklin (NHC - USA), Ed Fukada (JTWC - USA), T.C.Lee (HKO – Hong Kong, China), Buck Sampson (NRL - USA), Todd Smith (BoM - Australia) Abstract: Operational use of consensus track forecasting methods has become relatively widespread in the last four years. This paper gives an overview of current consensus track forecasting methods, summarises recent research regarding consensus track forecasting methods, and looks at implementation challenges faced by operational centers. We consider the relative merits and degree of operational use of single-model and multi-model ensembles, weighted and non-weighted consensus methods, selective and non-selective methods and use of a vector motion average in place of a geographical position average. The theoretical basis for optimising consensus forecast skill is considered in relation to operational issues that may create roadblocks to realising that skill. A number of recommendations are made to improve uptake of consensus methods amongst operational centers, particularly those with fewer resources, and to optimise track forecasting skill in operational centers already using consensus methods. 3a.1 Introduction The benefits of consensus forecasting have long been recognised by the meteorological community (Sanders 1973, Thompson 1977). A general subjective form of consensus track forecasting has been in widespread use ever since multiple forms of forecast track guidance have been available to forecasters. By the early 1990s, the potential for improved tropical cyclone track prediction from the objective blending of independent forecast tracks was explicitly recognised (Leslie and Fraedrich 1990), and a phase of experimentation and inconsistent operational use began (Jeffries and Fukada 2002). However, the chapter of “Global Guide to Tropical Cyclone Forecasting” on tropical cyclone motion (Holland 1993) makes no reference to consensus track forecasting. By the late 1990s, systematic use of objective consensus track forecasting methods (hereafter “consensus methods”) was established at the Joint Typhoon Warning Center (Jeffries and Fukada 2002). Since that time an increasing number of Tropical Cyclone Warning Centers (TCWCs) have adopted the consensus method. At the time of writing additional TCWCs known to have adopted the consensus method include the National Meteorological Center of China Meteorological Administration (NMC/CMA) ,Central Pacific Hurricane Center (CPHC), Hong Kong Observatory (HKO) (Lee and Wong 2002), Regional Specialised Meteorological Center (RSMC) Fiji, Australian Bureau of Meteorology (BoM) TCWCs (Perth, Darwin, and Brisbane), RSMC La Reunion, Vietnam National Center for HydroMeteorological Forecasting (NCHMF), and RSMC Tokyo Typhoon Center (JMA).

425

Clear evidence of the improvement in overall track forecasting skill obtainable via a consensus method has no doubt lead to its widespread uptake. However, work continues to find the optimal method for blending the available guidance. 3a.2 Consensus Track Forecasting Methods 3a.2.1 Interpolation of model tracks Forecasters are often required to make forecasts for times that are not available in the model output of some or all of the available guidance models. To maximise the availability of model track guidance, interpolation of the available guidance tracks is used to create interpolated model tracks with position forecasts for the required times. Investigation has shown that cubic spline interpolation offers no advantage over linear interpolation (Sampson et al. 2006a). All centers surveyed for this report were using linear interpolation. For guidance tracks available at 6-hourly resolution the differences should be minimal. When track guidance is only available at poor temporal resolution, interpolation can cause introduction of significant error. However, the requirement to interpolate model tracks to increase track guidance availability, while likely to have a small negative impact on the accuracy of the resulting track predictions, is offset by the skill gain afforded by the consensus method. 3a.2.2 Single and multi-model methods The generation of prediction ensembles for weather forecasting can be divided into two broad categories. First, a single numerical weather prediction (NWP) model can be integrated many times for the same base time using slightly different initial conditions. Most NWP centers now run such an Ensemble Prediction Systems (EPS). Although progress has been made in optimising perturbation methods in order to obtain a realistic spread in EPS model tracks (e.g., Puri et al. 2001), little work has been done on the long-term verification of the accuracy of an EPS mean track prediction against the deterministic track prediction of the same model NWP model, or that obtainable from a multi-model ensemble. The accuracy of NWP track prediction is known to be sensitive to model resolution and thus the requirement to run the EPS at a degraded resolution compared to the deterministic model run may impact on the accuracy of the ensemble mean track prediction. However, data from the annual NHC verification reports for the years 2001-2005 (Gross cited 2006, Franklin cited 2006) indicate that in the Atlantic and NE Pacific basins the GFS ensemble mean has similar average track prediction skill to 72 hours as the control run (Figure 3a.1). Most operational centers are not currently including the ensemble mean forecast as a member in their multi-model ensembles. An alternate approach to consensus track forecasting is to generate an ensemble prediction from a set of deterministic NWP models. This multi-model approach to consensus track forecasting has come to represent the current state-of-the-art in operational track forecasting. Consensus approaches using multi-model ensembles can be categorised as: 1) involving a weighted or non-weighted (hereafter “simple”) combination of guidance tracks; and 2) selective or non-selective. 3a.2.3 Weighted and non-weighted methods Goerss (2000) showed that a simple average of the position predictions of a number of numerical models could outperform any of the member models (when averaged on seasonal timescales). Earlier Leslie and Fraedrich (1990) had shown that a weighted linear combination of the track forecasts of two models (CLIPER and an NWP model) could outperform either of the member models. Vijaya Kumar et al. (2003) and Williford et al. (2003) describe the similar development of a weighted multi-model ensemble for track and intensity, described as a superensemble (hereafter FSSE – Florida State University Superensemble). The FSSE methodology applies unequal weights to each forecast parameter (for track forecasts: latitude and longitude) of each model for each forecast time. They demonstrate that the resulting weighted consensus can outperform the non-weighted average of the

426

same set of models. However, Williford et al. (2003, p. 1891) note, “When significant changes are made to the component forecast models, a sufficiently large, updated training set is required to capture the updated model characteristics. Otherwise, the superensemble may not outperform an ensemble mean of the better models.” Given the pace of NWP development, “retraining” (recalculation of the regression coefficients) is generally required at the start of each season and may also be required during the season, as was the case for the 2000 Atlantic season (Williford et al. 2003). This maintenance overhead may be an impediment to implementation in many operational centers. In contrast, the ease of implementation and lack of maintenance overhead of the Goerss’ method has resulted in it becoming the baseline forecasting method in many operational centers including BoM, HKO, JTWC, NCHMF, NHC and RSMC La Reunion. Weber (2004) has developed a probabilistic consensus method (Probabilistic Ensemble System for the Prediction of Tropical Cyclones (PEST)) that is capable of producing both deterministic forecast positions and geographical strike probability maps. One interesting aspect of Weber’s approach was to include all available guidance including other consensus predictions as ensemble members. When verified on global TC data for 2001 and 2002, the PEST mean annual (“deterministic”) forecast position errors were comparable in quality to that of other consensus methods. Weber (2004) argues that modification and updates to individual members of the ensemble, or addition of new members, has little consequence for PEST, which gives it an advantage over the weighted consensus method of Vijaya Kumar et al. (2003). However, the lack of improvement over the simpler approach of an unweighted consensus such as that of Goerss (2000) does not encourage operational implementation. The potential for PEST to produce multiple “deterministic” forecast positions (because the deterministic forecast positions are local maxima in the geographical probability distribution) may also be seen as an undesirable attribute in operational centers. In section 3a.4 Track Forecast Confidence – Guidance on Guidance, we discuss the growing demand for objective measures of confidence in track forecasts, and it is in this area that the greatest operational value of PEST may lay.

Figure 3a.1. Homogeneous comparison of average model track error for the GFS control (GFS) and GFS ensemble mean (GEMN) in the North Atlantic and North East Pacific basins over the period 2001-2005. The number of cases is shown in parentheses.

427

Figure 3a.2 (from Sampson et al. 2006) Performance of the selective consensus for all SAFA analysis cases in the years 2000-2004. (a) 72-h forecast improvement (percent) of selective consensus over non-selective consensus. The number of cases is shown in parentheses. (b) Percentage of total forecasts for which a selective consensus was produced. The total number of SAFA analyses and selective consensus forecasts, respectively, are shown in parentheses. 3a.2.4 Selective and non-selective methods The weighting of individual forecasts based on past performance is one potential method of improving on the simple, non-weighted consensus method. The other approach that has received considerable attention, and has been systematically evaluated, is the selective consensus method. The Systematic Approach Forecast Aid (SAFA) is a knowledge-based tropical cyclone track forecast system developed to assist forecasters in the information management, visualisation and proactive investigation of error mechanisms associated with track prediction in NWP models (Carr et al. 2001). A key component of SAFA is the construction of a selective consensus (SCON) based on the exclusion of NWP model tracks suspected of having a 72-h forecast position error greater than 300 n mi (1 n mi = 1.85 km). Although a suspected 300 n mi forecast error is a necessary condition to eliminate a NWP model forecast, a large spread (outlier greater than 250 n mi from the position of the non-selective consensus of the SAFA models (NCON)) and an error mechanism must also be present. SAFA was installed for operational evaluation at JTWC in 2000 and evaluation continued through 2004 (Sampson et al. 2006a). Sampson et al. (2006a) found that over the evaluation period the JTWC forecasters produced fewer (~ 5%) SCON forecasts, so that combined with the 95% of other forecasts no statistically significant improvement on the seasonal forecast errors was found. In its first year of evaluation, SCON suffered from overuse (creating SCON when it was not justified) and the overall performance in this first season indicated the 72-h SCON forecasts were degraded by 7.5% with respect to NCON forecasts (Sampson et al. 2006a). A need for improved training of forecasters in the use of SCON was identified (Jeffries and Fukada 2002). This experience highlights one of the principal advantages of non-selective consensus methods: application of non-selective methods requires little skill and therefore minimal training. In operational centers where high levels of training are difficult to support, this is a significant consideration. With additional training and restriction on its use, the number of SCON forecasts dropped markedly the following year and the quality of the SCON forecasts improved (Figure 3a.2, from Sampson et al. 2006).

428

Although on average SCON outperformed NCON for those few cases that an SCON was formed, over the remainder of the evaluation period, the positive impacts on the overall seasonal errors were not statistically significant (ibid). SCON use continued to decline over the evaluation period to the point of making the approach irrelevant (Figure 3a.2), partly as a result of model improvements leading to less opportunity for SCON creation (ibid). Additional factors included the creation of new objective consensus aids that were essentially non-selective consensus tracks constructed from a larger pool of NWP models than that incorporated into SAFA. These new consensus forecasting aids were known to perform better than the equivalent non-selective consensus of the models incorporated into SAFA (NCON) (Goerss et al. 2004). During the 2005 season, SCON production was dropped from the JTWC operations. This highlights one of the other advantages of non-selective, consensus methods over selective (and weighted) consensus methods: non-selective, non-weighted consensus methods have minimum maintenance overheads and can incorporate new models “on-the-fly.” On the other hand, consensus methods that require ongoing maintenance (in this case the work required to incorporate new models into SAFA), are disadvantaged in an operational setting. In TCWCs other than the JTWC, SAFA has not been rigorously implemented. The NHC tested a version of SAFA (renamed Dynamical Model Track Prediction Evaluation System (DYMES)) and declined to implement on the basis that it was rarely successful in improving over a non-selective consensus and that it was time-consuming to use. Although SCON, the specific form of selective consensus produced under SAFA, was not widely implemented and is no longer operationally produced, other selective consensus methods are widely used in TCWCs. Where there is good agreement between the model tracks, the forecaster has no imperative to exclude models and will have greater confidence in the track forecast. A forecast based on an ensemble with a small spread - although not guaranteeing low error - does provides greater confidence in an accurate track prediction. Forecasts from an ensemble with large spread may be either poor or good (Aberson 2001, Buizza and Palmer 1998). Where there is generally good agreement between the models, but one or two models present as extreme outliers, a forecaster is more likely to exclude the outliers and opt for the consensus of the remaining models. At the HKO, forecasters may use space mean analysis (Bell and Lam 1980; Dong and Neumann 1986), cluster analysis of track forecasts and strike probability maps generated from ECMWF and JMA EPS are used to determine which (if any) model tracks to discard. In some cases the exclusion of a model is made along SAFA-lines (i.e., based on knowledge of individual model biases). For example, Perth TCWC forecasters have observed a pronounced left-of-track bias in the Australian Tropical Cyclone Limited Area Prediction Scheme (TCLAPS) during recent seasons. One trap with all such “knowledge-based” approaches (including SAFA) lies in the time lag between the changes in individual model bias and the updating of the knowledge base. In addition to changes in individual model bias, the relative performance of NWP models changes with time. Hence just as weighted consensus methods require “retraining,” forecasters who subjectively weight their forecasts toward models with greater average historical skill face a similar trap. Moreover, forecasters may be influenced by short-term trends in model performance (recent excellent track predictions from a particular model) that are not indicative of longer-term trends. Tropical cyclone forecasters may also exclude an NWP model where the short-term forecast fields has a marked deviation from the current (manual) analysis of either the tropical cyclone or the broader environment. This is possible because NWP model runs become available to forecasters some hours after the nominal analysis time and each model run is typically used for another six to twelve hours before a more recent model run becomes available.

429

A detailed diagnosis of model fields is by no means an easy task and the difficulty for the forecaster lies in qualitatively determining how far a model must be “off-track” before it should be excluded from the consensus. A model that may be diagnosed as having a poorer initial representation may turn out to have the more accurate track prediction, as the following case illustrates. Intense Tropical Cyclone Francesca formed in the southwest Indian Ocean (SWIO) during early February 2002. The 1200 UTC 3 February 2002 analyses of the United Kingdom Meteorological Office global model (UKMO) and the European Center for Medium-range Weather Forecasts (ECMWF) model were very similar in their depiction of the amplitude and position of synoptic features across the SWIO (not shown). The only discernible difference between the two was that the UKMO, which is a “bogussed” model, had a better initial representation of the cyclone position. Nevertheless, the 42-h ECMWF cyclone position forecast proved much more accurate (within 93 km) compared with the UKMO 42-h forecast error of 440 km. Interestingly, streamlines of the 500 hPa and 400 hPa 42-h forecast fields of both models are strikingly similar, with respect to the relative position and amplitude of synoptic features (Caroff et al. 2004). Hence we have an example where both models similarly analyse the situation and its evolution and yet they result in significantly different track forecasts. Despite the difficulties and risks involved in the subjective application of selective consensus methods, the potential still exists for forecasters to add considerable value to track predictions in specific situations. Seasonal verification statistics do not reflect occasional instances of very large errors. These cases are of great interest to operational centers not just because of the magnitude of the error but also because of the disproportionate effect they have on user confidence in track predictions. It is these instances where forecasters have the greatest potential to add skill to track prediction. Such instances may not all be associated with large ensemble spread. Research that targets the application of consensus methods in cases where large errors have resulted may be required for progress in this critical area. 3a.2.5 Position consensus vs vector motion consensus During the 2005-2006 Southern Hemisphere (SH) tropical cyclone season, BoM introduced a change to their consensus method. Previously the BoM consensus method was similar to that described by Goerss (1998) in which the forecast positions of a set of objective aids are geographically averaged to determine the consensus forecast position. Similar to the HKO approach, the BoM consensus method involves construction of an “on-the-fly” consensus in which all models available at the forecast time is required are utilised, so the number of consensus members varies according to the availability. Originally the BoM consensus method required that all forecast hours of a given consensus have homogeneous membership, which effectively truncated the consensus forecast to the shortest lead-time of any member. This method can lead to consensus forecasts that are shorter than the operational requirement and hence to the subjective manipulation of longer lead-time forecast positions. To redress this, the BoM method was amended to allow inhomogeneous membership with forecast lead-time, and the forecast positions were displayed in a manner that allowed forecasters to identify how many members had contributed to a given forecast position. It was observed that combining short- and medium-range models in this manner could give rise to a discrete change in the consensus forecast position as the number of objective aids was reduced at longer lead times (Figure 3a.3b). It was noted that when an initial position correction is being applied to the objective aids it is possible to construct a consensus track forecast based on the average vector motion of the objective aids rather than the average geographical position. When the number of consensus members does not vary with lead-time, the two approaches give the same result. The vector motion average method has the advantage of always providing a smooth forecast track (Figure 3a.3c). This approach has also been considered (but not yet tested) in relation to weighted consensus methods by Vijaya Kumar et al. (2003).

430

It is common for an initial position correction to be applied to model tracks before they are used to construct a consensus track. If the position correction is small then this should cause a slight increase in skill. However, if the model position is a significant distance from the analysed position, then it can be argued that the environmental steering flow represented in the model may be significantly different from that in the vicinity of the cyclone and it would be invalid to include the model in the consensus. For this reason, models with gross analysis errors are occasionally excluded from a consensus method by some centers (e.g., BoM). Other centers (e.g., NHC) do not exclude models on this basis. During cyclogenesis it can be difficult to accurately identify the position of the seedling disturbance when a coherent long-lived low-level circulation center (LLCC) is not yet well defined. Tropical cyclones often form from a Mesoscale Convective System (MCS) in which the vorticity is maximized in the mid-troposphere and decreases above and below. In the Australian region, it has been observed that when a MCS is under steady translation in an environment of moderate vertical wind shear, from time to time well-defined small-scale LLCCs will become evident in the satellite imagery up-shear of the MCS. If the vertical shear is maintained, as the MCS translates farther west the LLCC generally becomes increasingly separated from the convection and new LLCCs are observed in the “wake” of the MCS. The failure to identify this process can lead to discontinuous shifts in the analysed position of the seedling disturbance. A similar problem occurs with cyclogenesis in the monsoon trough where multiple small-scale LLCCs are commonly observed within a broad area of weaker winds. One of these centers may eventually become a coherent long-lived LLCC, or a new LLCC may develop. At this stage of development, it is often difficult to accurately determine the LLCC. It can be argued that there is no skill gain to be had from performing an initial position correction when the initial position cannot be accurately analysed. Whereas it is not possible to create a vector-motion consensus forecast without performing an initial position correction, it is still possible to generate a geographical consensus forecast. Consequently BoM TCWCs have recently adopted the approach of using a vector motion average whenever an initial position is being applied, and using a geographical position average when the initial position cannot be confidently analysed.

431

Figure 3a.3. Consensus forecast tracks generated from the tracks shown in (a) (with initial position correction applied) using (b) average geographical position and (c) average vector motion.

432

3a.3 Optimising Consensus Track Forecasting The average forecast error of an unweighted consensus forecast has been shown to be dependent on three things: 1) the number of models included in the consensus (Goerss et al. 2004, Buizza and Palmer 1997, Leith 1974); 2) the mean forecast error of the individual models that constitute the consensus Goerss (2000a); and 3) the degree of independence (or effective degrees of freedom) of the forecast errors of the individual models (ibid). 3a.3.1 Effect of ensemble size on accuracy In a study with an EPS, Buizza and Palmer (1997) showed that the skill of the ensemble mean was dependent on ensemble size only when up to eight members were included in the ensemble. This is a ‘law of diminishing returns’ and is expected to apply, at least qualitatively, to multi-model ensembles as well. While forecasters readily add members to the consensus that are considered to have individual skill, there is often a reluctance to add objective aids that are less skillful (BoM and NHC forecasters, personal communication). This concern is naturally exacerbated if the number of available guidance tracks is low (and hence the less skillful model will exert a greater influence on the resulting consensus). Krishnamurti et al. (2000) give some weight to such concerns, noting that: “The (unweighted) ensemble average appeared to degrade in skill if more and more models with lower skills were used in the averaging process.” Goerss (2000) considered a pool of three models where one of the models had a significantly poorer skill level than the other two models. A homogeneous comparison of the three-model consensus against the consensus of the best two models showed that omission of the model with less average skill did not improve the forecast, in fact a slight but not statistically significant degradation of skill occurred. 3a.3.2 Independence of consensus members Sampson et al. (2006) demonstrate that a model with relatively lower forecast skill (but still with better skill than CLIPER) can have one of the largest positive impacts on the skill of an unweighted consensus when that model has greater independence from the other members of the consensus. Vijaya Kumar et al. (2003) show that a weighted consensus method can significantly improve on the skill of the individual models even where the models are highly correlated (low independence). In their results, a strong correlation between the members of the consensus resulted in the unweighted ensemble mean being only slightly better than the individual models. However, the weighted consensus track proved significantly more skilful. 3a.3.3 Optimal consensus membership Given the dependence of consensus accuracy on the three factors listed above, the question arises as to how TC forecasters can determine when the addition of a model of a given individual skill level will provide a positive contribution to the average accuracy of an unweighted consensus. Skill is generally defined as the ability to improve upon simple techniques such as those based on climatology and persistence, and so it has become common practice to measure the track forecasting skill of objective aids against that of the Climatology and Persistence forecast (CLIPER; Neumann 1972). It could therefore be argued that a simple benchmark for inclusion of a model in a consensus is that it has greater average skill than CLIPER. However if we consider a consensus already formed from a reasonably large set (say N>>8) of skilled guidance tracks exhibiting significant independence, then the above discussion suggests that if we were to add a significant number of related (non-independent) guidance tracks, each only marginally outperforming CLIPER, accuracy of the resulting consensus may

433

be degraded. The evidence presented by Goerss (2000) and Sampson et al. (2006) clearly supports the methodology of the JTWC and the HKO where models are included in the consensus based not on their individual skill but on a comparison of the skill of the consensus formed with and without the candidate model. For NWSs with access to a small number of NWP tracks, the work of Buizza and Palmer (1997) suggests that the focus should be on gaining greater access to model tracks. In their implementation of consensus track forecasting, the Vietnam NCHMF has taken advantage of the multiple advisories available for their area of interest by using the operational track forecasts of the JMA, the JTWC, and the Chinese Meteorological Agency (CMA) as consensus members. 3a.4 Track Forecast Confidence - Guidance on Guidance The level of confidence that can be assigned to a specific track forecast (cf. mean skill verified over a given historical period) is a parameter of intrinsic value to risk managers. Forecasters often view the spread of the multi-model consensus as well as that of single model EPS to gauge a level of confidence in the track forecast. As previously mentioned (section 3a.2.3), forecasts from an ensemble with large spread may be either poor or good (Aberson 2001, Buizza and Palmer 1998). Hence it is not straightforward to determine track forecast confidence from ensemble spread. PEST provides a method of directly generating strike-probability maps that may be of significant value to risk managers. As discussed in section 3a.2.3, the possibility of multiple local maxima in strike probability, while arguably a benefit of this approach, may detract from its suitability to operations where the emphasis remains on a single deterministic track forecast. Goerss (2004, 2006) has developed a method for the prediction of consensus error (GPCE) that is applicable to simple consensus methods in common usage. Stepwise linear regression of a pool of predictors available prior to the forecast deadline was used to determine to what extent the TC track forecast error of an unweighted consensus could be predicted. For the 2001-2003 Atlantic seasons, the most important predictors were found to be consensus model spread and TC intensity (either initial or forecast). The regression models were able to explain 15-20% of the track forecast error variance for shorter forecast lengths (24-72 h) and 45-50% of the track forecast error variance for longer forecast lengths (96-120 h). By adding a constant (varied with respect to forecast length), radii were derived to draw circular areas around each consensus forecast position. The additive constants were chosen so that the verifying TC position was contained within the circular area surrounding the consensus forecast position 72-74% of the time for the 2001-2003 Atlantic seasons. These predicted radii varied from approximately 30-140 n mi at 24 h, 55-260 n mi at 48 h, 65-550 n mi at 72 h, 90-1000 n mi at 96 h, and 125-1175 n mi at 120 h (ibid.). Based on the size of these radii, a forecaster can determine how much (or little) confidence can be ascribed to the consensus forecast position. Two contrasting examples of GPCE output forecast are shown in Figure 3a.4a-b. Independent data tests of the GPCE produced results that compared quite favourably with those from dependent testing, which suggests that the regression model developed from 2001-2003 data could be applied to the 2004 TC season. Further work is underway to extend the approach to other basins, optimise the length of training period, investigate the use of more sophisticated regression methods, and determine the impact of dataset stratification (based principally on recurvature) on determination of the regression models (ibid.).

434

Figure 3a.4 (from Goerss 2006) Predicted 72% confidence radius (solid) surrounding the 120-h CONU forecast for Hurricane Kate from (a) 00 UTC 30 September 2003 and (b) 00 UTC 13 September 2003. The individual model tracks used to create the CONU track are shown along with the 120-h radius (dotted) used in the NHC Potential Day 1-5 Track Area graphic.

435

3a.5 Roadblocks and recommendations 3a.5.1 Access to model tracks One of the biggest roadblocks to wider uptake of consensus track forecasting techniques is the availability of NWP guidance tracks. The significance of this roadblock is likely to be greatest for operational centers with minimal resources. At present there is no central repository of NWP guidance tracks and no agreed format for their exchange. NHC and JTWC run a number of vortex trackers and likely have access to the greatest number of NWP tracks. All model tracks are automatically ingested at these centers (i.e., there is no need for forecasters to manually input NWP tracks). At the HKO, the NWP forecast tracks disseminated by UKMO and JMA are ingested into software (the Tropical Cyclone Information Processing System (TIPS)) developed to generate and display the consensus track on-the-fly (Tai and Ginn 2001). The TC forecast positions of the ECMWF and NCEP models are determined from the surface prognoses as the point of minimum mean sea-level pressure, which is identified by overlapping parabolic interpolation (Manning and Haagenson 1992). An interactive tool is used to extract these forecast positions for the TIPS. The RSMC La Reunion runs a vortex tracker on several models and supplements this with the output of vortex trackers from other centres. However delays in receiving the output from some centres frequently restricts the number of models available for the consensus and one of their major ongoing efforts is aimed at enlarging the number of consensus members via an increase in the number of models and an improvement in the timeliness of their arrival. The BoM runs a vortex tracker on the Tropical Cyclone Limited Area Prediction Scheme (TCLAPS) model, but otherwise relies on access to the output of vortex trackers run by the JTWC, the JMA, the ECMWF and the UKMO, which are supplemented with the manual input of guidance tracks from model parameter displays. The manual input of NWP guidance tracks is undesirable for two reasons. First, it introduces an element of subjectivity, especially when the displayed resolution of the model fields is poor. Of greater concern from an operational point of view is the time-consuming nature of the process, which acts to limit the number of members in the consensus (insufficient time to enter the tracks) and/or reduce the time available to the forecaster for analysis and forecasting. This has been identified as a significant operational problem at BoM TCWCs. Many smaller National Weather Services (NWS) have access to only one or two NWP models. Understandably these nations often desire a degree of independence in tropical cyclone forecasting, which occasionally leads to situations where warning policy is weighted toward the track prediction of a single (locally available) model over the generally superior guidance provided by the relevant RSMC using consensus methods. The broad and free dissemination of NWP guidance tracks, preferably in a standard format, is required for widespread uptake of consensus track forecasting methods. NWSs with fewer resources are likely to have less access to NWP guidance tracks and this imbalance may impact on disaster mitigation in less developed nations. To allow RSMCs/TCWCs to access the latest TC track forecasts from different global /regional models in an efficient and timely manner, it is desirable to establish a “one-stop-shop” website similar to that of the Severe Weather Information Center (SWIC) website (http://severe.worldweather.wmo.int/) of the WMO for bringing together the latest TC guidance / forecasting tools from different NWP centers for operational TC forecasting. The "Numerical Tropical Cyclone Prediction Web Site" (https://tynwp-web.kishou.go.jp – password protected) implemented by JMA for displaying TC forecast tracks provided by major NWP centers in graphical form for the members of the ESCAP/WMO Typhoon Committee is certainly a good example and encouraging start (Kyouda 2006).

436

3a.5.2. Need for interpolation and initial position correction Forecasters are required to generate forecasts for specific times and those times may vary across operational centers. Thus access to software capable of interpolating the model tracks to the times of interest is generally necessary for the implementation of consensus methods. There may also be the requirement for the software to be able to perform an initial position correction to the model tracks before interpolation. Although this is not a difficult technical challenge and for most operational centers this is not a significant roadblock, it may be a consideration for smaller operational centers. With respect to interpolation, the most practical assistance that could be offered may be to ensure that NWP tracks are interpolated to hourly positions prior to dissemination, which removes the requirement for further interpolation. Initial position correction can only be performed following analysis of the cyclone position and hence supporting this element of consensus forecasting methodology requires provision of access to suitable software. This correction might be achieved through projects that encourage major RSMCs/TCWCs to share basic software tools with smaller NWSs in the region. However, provision of this software through a web interface associated with the “one-stop-shop” referred to above would facilitate the widest possible access. 3a.5.3. Access to model fields for diagnosis TCWCs may have access to the forecast track of an individual NWP model, but not to the forecast fields. This limits the forecaster to a decision to include or exclude the model track based on gross characteristics of the track. While some of the results presented in recent years suggest that forecasters should indeed limit themselves to excluding model tracks with gross errors, the development of a mental picture of different forecast scenarios remains an important part of the operational forecasting process. Knowledge of the range and relative likelihood of various forecast scenarios improves the products and briefings provided by the forecaster and thereby adds considerable value to the warning service. It is not essential for the forecaster to have access to the model fields of every ensemble member, particularly since operational constraints are unlikely to allow a detailed diagnosis of every model. However, it is desirable for the forecaster to have access to a suite of models representative of the range of possible outcomes. 3a.5.4. Training needs Adequate training is essential to the successful operational implementation of any new forecasting technique, and the experiences of the JTWC and the BoM have shown that this is true for consensus forecasting methods. Other experiences at the BoM have shown that forecasters are more likely to use new forms of objective guidance if they are presented with clear evidence of improved skill and are able to grasp the scientific basis for the increase in skill. Conversely forecasters tend to resist the implementation of “black box” techniques. Forecasters are generally well adapted to a form of subjectively weighted and subjectively selective track forecasting. Hence one aspect of the implementation of unweighted consensus methods that may require focused training is the idea that an individual model with lesser (greater) skill but greater (lesser) independence can add more (less) skill. It is recommended that operational centers adopting consensus track forecasting methods provide appropriate training to forecasters to ensure successful implementation. Regional assistance will help smaller operational centers learn from the experience of major RSMCs/TCWCs in using consensus methods and EPS products. With weighted consensus methods, a different kind of training is required that could prove a roadblock to successful long-term implementation. As previously discussed, the changing nature of NWP models requires that weighted consensus methods are continually “retrained” and this maintenance overhead may prove to be a roadblock for continued use. Similarly selective methods such as SAFA that are

437

based on knowledge of systematic biases and error mechanisms in NWP models involve a retraining requirement that may be a hurdle to operational use. It is recommended that operational centers give close consideration to the maintenance overheads associated with any potential consensus methodology before implementation. 3a.5.5. Bifurcation scenarios Occasionally the track of a cyclone is sensitively dependent on minor changes in the interaction of the tropical cyclone with a mid-latitude trough system. The most common example is where the recurvature of a system is sensitively dependent on the amplitude and timing of the interaction with a mid-latitude trough. The output from the 00 UTC 12 August 2004 run of the ECMWF EPS for Hurricane Charley in Figure 3a.5 demonstrates this situation. Continued improvement in NWP may eventually establish levels of skill in deterministic track forecasting that make such difficult forecasting decisions extremely rare. In the meantime, while the service requirement for a single deterministic track forecast is maintained, it is difficult to see how the onerous position of the forecaster in this circumstance can be improved.

438

Figure 3a.5. ECMWF EPS probability distribution for Hurricane Charley from 00 UTC 12 August 2004 that has a forecast track probability distribution with a bifurcation involving either a recurvature or a missed-recurvature.

Figure 3a.6. (from Franklin, cited 2006b). Homogenous comparison for selected Atlantic basin early intensity guidance models for 2005.

439

3a.5.6. Consensus methods for intensity forecasting

Skill in intensity forecasting has failed to keep pace with track forecasting skill. Encouraged by the success of consensus track forecasting, an increasing amount of research is being conducted into consensus methodology to intensity forecasting, with some success already recorded. Verification for the 2005 Atlantic season indicates that a two-member consensus intensity forecast created from the GFDL hurricane model and Decay SHIPS model, both skilful models, provided better operational guidance than either model alone. Indeed this simple consensus outperformed the FSSE intensity forecast despite the use of the previous official forecast as input to the FSSE (Figure 3a.6). Sampson et al. (2006c) used a statistical-dynamical model (STIPS) with the predictors derived from the outputs of ten NWP models to form a simple consensus intensity forecast. Some of the fields required for STIPS were not available for some of the models, which was addressed by substituting fields from another NWP model (e.g., NOGAPS), although this may have affected both the skill and independence of the consensus members. Preliminary tests with a small dataset showed promise. The gain in skill from forming an intensity consensus was generally not as large as the gain in skill from forming a track consensus of the same ten members. This result was anticipated, as the skill and independence of the consensus member intensity forecasts are less than the skill and independence of the track forecasts. Long-term plans include gaining access to field data for all consensus members and acquiring other independent skilful intensity forecasts. The Naval Research Laboratory Monterey (NRL) has commenced a project to investigate consensus intensity forecasting, with a focus on cyclogenesis, by applying multivariate regression to a set of forecast parameters from the available NWP models. This is similar to the process used by Sampson et al. (2006c) without the explicit formulation of a STIPS model. Initial results have been encouraging (Grant Elliott, personal communication). 3a.6 Summary Use of consensus track forecasting methods provides additional forecasting skill over that obtainable from any individual model. Clear evidence for this has lead to wider implementation of consensus methods. A number of variations in consensus methods have been examined and the relative merits of each approach have been discussed. Access to a number of NWP track predictions and adding models, up to at least 8 models, can increase skill. Access to model tracks is the most significant technical roadblock to implementation of consensus methods at additional operational centers, and it is recommended that steps be taken to improve open access to NWP model tracks in a standard format. Experience has shown that some degree of automation and software support is essential to effective implementation and it is recommended that consideration be given to ways in which NWSs with fewer resources can be supported in this area. Experience has also shown that appropriate training is critical to uptake and appropriate use of consensus methods. To this end it is recommended that WMO tropical cyclone training courses include specific training on the use of consensus approaches and that operational centers looking to implement consensus track forecasting methods include targeted training as a part of their implementation plan. Despite continued overall improvement in track prediction of NWP models, instances of large track prediction errors still arise, even following the application of consensus methods. Research that improves the application of consensus methods in cases where large errors have resulted may be needed for progress in this critical area.

440

References

Aberson, S., 2001: The ensemble of tropical cyclone track forecasting models in the North Atlantic basin (1976-2000). Bull. Amer. Meteor. Soc., 82, 1895-1904.

Buizza, R., and T. N. Palmer, 1998: Impact of ensemble size on ensemble prediction. Mon. Wea. Rev., 126, 2503–2518.

Bell, G.J., and Lam, C.Y., 1980 : Departures of tropical cyclone movement from geostrophic steering. Proceedings of the International Symposium on Typhoons, Shanghai, 6-11 October.

Caroff, P., P. Aliaga, C. Bientz, A. Charlat, A-C. Fontan, and M. Mayoka, 2004: Cyclone season South-West Indian Ocean 2001-2002. [source]

Carr, L. E. III, R. L. Elsberry, and J. E. Peak, 2001: Beta test of the systematic approach expert system prototype as a tropical cyclone forecasting aid. Wea. Forecasting, 16, 355-368.

Dong, K., and C. J. Neumann, 1986: The relationship between tropical cyclone motion and environmental geostrophic flows. Mon. Wea. Rev., 114, 115-122.

Franklin, J. L., cited 2006a: 2005 National Hurricane Center forecast verification report. [available online at http://www.nhc.noaa.gov/verification/pdfs/Verification_2004.pdf]

_______, cited 2006b: 2004 National Hurricane Center forecast verification report. [available online at http://www.nhc.noaa.gov/verification/pdfs/Verification_2005.pdf]

Goerss, J., 1998: Global model tropical cyclone forecast performance. Minutes of the 52nd Interdepartmental Hurricane Conference, 26-30 January 1998. Clearwater Beach FL, Office of Federal Coordinator for Meteorological Services and Supporting Research, A61-62.

_______, 1999: Tropical cyclone forecasting using an ensemble of dynamical models: 1998 Atlantic hurricane season. Preprints, 23rd Conf. Hurr. Trop. Meteor, Dallas,TX, Amer. Meteor. Soc., 826-827.

_______, 2000a: Tropical cyclone track forecasts using an ensemble of dynamical models. Mon Wea. Rev., 128, 1187-1193.

_______, 2000b: Quantifying tropical cyclone forecast uncertainty using an ensemble of dynamical models. Preprints, 24th Conf. Hurr. Trop. Meteor., Ft. Lauderdale, FL, Amer. Meteor. Soc., 429-430.

_______, 2006: Prediction of consensus tropical cyclone track forecast error. Submitted to Mon Wea. Rev.

_______, C. R. Sampson, and J. M. Gross, 2004: A history of western North Pacific tropical cyclone track forecast skill. Wea. Forecasting, 19, 633–638.

Gross, J. M., cited 2006a: 2001 National Hurricane Center forecast verification report. [available on line at http://www.nhc.noaa.gov/ verification/pdfs/Verification_2001.pdf]

_______, cited 2006b: 2002 National Hurricane Center forecast verification report. [available online at http://www.nhc.noaa.gov/ verification/pdfs/Verification_2002.pdf]

_______, cited 2006c: 2003 National Hurricane Center forecast verification report. [available online at http://www.nhc.noaa.gov/ verification/pdfs/Verification_2003.pdf]

Holland, G.J., 1993: Tropical cyclone motion. Chapter 3, Global Guide to Tropical Cyclone Forecasting, G. J. Holland, Ed., WMO/TD-560, World Meteorological Organization, Geneva, Switzerland.

Jeffries, R. A., and E.J. Fukada, 2002: Consensus approach to track forecasting. Paper TP3.2, Extended Abstracts, Fifth International Workshop on Tropical Cyclones, Cairns, Australia, World Meteorological Organization (Geneva).

441

Krishnamurti T.N., T.S.V. Kumar, Z. Zhang, T. LaRow, D.R. Bachiochi, C.E. Williford, S. Gadgil and S. Surendran, 2000: Multi-model superensemble forecasts for weather and seasonal climate. Proceedings of the 24th Conference on Hurricanes and Tropical Meteorology, Fort Lauderdale, FL, Amer. Meteor. Soc.

Kyouda, M., 2006: Report on applications of EPS for severe weather forecasting. WMO Commission of Basic Systems, OPAG DPFS, Expert meeting on Ensemble Prediction Systems, Exeter, UK, 6-10 February 2006

Lee, T. C., and M. S. Wong, 2002: The use of multiple-model ensemble techniques for tropical cyclone track forecast at the Hong Kong Observatory. WMO technical conference on data processing and forecasting systems, Cairns, Australia, 2-3 Dec. 2002.

Leith, C. E., 1974: Theoretical skill of Monte Carlo forecasts. Mon. Wea. Rev., 102, 409–418.

Leslie, L.M., and K. Fraedrich, 1990: Reduction of tropical cyclone position errors using an optimal combination of independent forecasts. Wea. Forecasting, 5, 158-161.

Manning, K.W., and P.L. Haagenson, 1992: Data ingest and objective analysis for the PSU/NCAR modeling system: Programs DATAGRID and RAWINS. NCAR Technical Note, NCAR/TN-376+IA, 209 pp.

Neumann, C., 1972: An alternate to the HURRAN tropical cyclone forecast system. NOAA Tech. Memo. NWS SR-62, 24 pp.

Sampson, C.R., J.A. Knaff, and E.M. Fukada, 2006a: Operational evaluation of a selective consensus in the western North Pacific basin. Wea. Forecasting, Submitted

_______, J.S. Goerss, and H.C. Weber, 2006b: Operational performance of a new barotropic model (WBAR) in the western North Pacific basin. Wea. Forecasting, in press.

_______, J.A. Knaff, and M. DeMaria , 2006c: A statistical intensity model consensus for the Joint Typhoon Warning Center. 27th Conference on Hurricanes and Tropical Meteorology, 24-28 April, Monterey, CA, Amer. Meteor. Soc. [available online at http://ams.confex.com/ams/pdfpapers/107554.pdf]

Tai, S.C., and W. L. Ginn, 2001: Tropical Cyclone Processing Systems (TIPS) of the Hong Kong Observatory, Eighth Workshop on Meteorological Operational Systems, ECMWF, 12-16 Nov 2001

Thompson, P.D., 1977: How to improve accuracy by combining independent forecasts. Mon. Wea. Rev., 105, 228-229.

Vijaya Kumar, T. S. V., T. N. Krishnamurti, M. Fiorino, and M. Nagata, 2003: Multimodel superensemble forecasting of tropical cyclones in the Pacific. Mon. Wea. Rev., 131, 574–583.

Weber, H.C., 2005: Probabilistic prediction of tropical cyclones. Part I: Position. Mon. Wea. Rev., 133, 1840–1852.

Williford, C. E., T. N. Krishnamurti, R. Correa-Torres, S. Cocke, Z. Christidis, and T. S. V. V. Kumar, 2003: Real-time multimodel superensemble forecasts of Atlantic tropical systems of 1999. Mon. Wea. Rev., 131, 1878–1894.

442

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES Topic 4 : Climate Variability and Seasonal Prediction of Tropical Cyclone Activity/Intensity Topic Chair: Dr. Chris Landsea NOAA/NWS/NCEP/TPC/National Hurricane Center 11691 SW 17th Street Miami, Florida 33165-2149, USA Email: Chris.Landsea@noaa.gov Phone: 1-305-229-4446 Fax: 1-305-553-1901 Incorporating: Topic 4.1 Variability of tropical cyclone activity/intensity on intraseasonal and interannual scale Rapporteur: Chang-Hoi Ho (SNU, South Korea) Topic 4.2 Possible relationships between climate change and tropical cyclone activity Rapporteur: Tom R. Knutson (GFDL, USA) Topic 4.3 Short-term climate (seasonal and intra-seasonal) prediction of tropical cyclone activity and intensity Rapporteur: Suzana J. Camargo (IRI, USA) 4.0.1: Introduction In this section, an overview of variability of tropical cyclone activity on timescales from intraseasonal and interannual in Topic 4.1 to multidecadal and millennial in Topic 4.2 is given. Predictions on the intraseasonal and interannual time frames are the focus of Topic 4.3. Additionally, a review of how anthropogenic greenhouse gas increases are impacting tropical cyclones today and in the future is discussed Topic 4.2. Previously, climate variability of tropical cyclones on scales from intraseasonal to millennial time frames was not in the mainstream of tropical cyclone research. However, interest in better understanding and prediction of relatively short-term (intraseasonal and interannual) fluctuations as well as longer (decadal, multidecadal, centennial, and millennial) variations has grown tremendously in the past several years. Additionally, there has been an explosion in publications and research into how the impacts of manmade “global warming” may be changing tropical cyclone characteristics both today and decades into the future. It is estimated that the tropical cyclone climate variability and change publications have increased by a factor of five to ten compared with just a decade ago, as the extensive literature review in these three topics demonstrate. While there has been an amazing growth within the field, there are numerous questions that remain which need to addressed by further research, such as: 1) Is there sizable and useable skill present for intraseasonal tropical cyclogenesis on the scale of a few days to a few weeks in advance?; 2) Can seasonal tropical cyclone predictions be regionalized with skill over and above just downscaling basin-wide predictions?; 3) What role, if any, does the stratospheric quasi-biennial oscillation have for dictating tropical cyclone activity and, if so, what is the physics involved?; 4) Given the climate surprises that occur on an interannual timescale (e.g., the unpredicted El Niño of 2006), what is the practical limit of predictability of seasonal forecasts?; 5) Are there naturally occurring significant multidecadal tropical cyclone oscillations?; 6) What are the trends in tropical cyclone frequencies and

443

intensities, after accounting for increasing monitoring during the last few decades?; 7) Given the wide disparity between global warming impacts on numerical modeling/theory (small impact on tropical cyclone intensity in the future) versus recent observational studies (large impact on tropical cyclone intensity already occurring today), which of these two conclusions is correct or are both in need of revision?; 8) What can tropical cyclone historical archival reconstructions and paleotempestology tell us about today’s climate and how changes may be manifested because of global warming and natural climate variability? All of these important issues are currently open ones that further study in theoretical, numerical modeling, and observational studies are needed to provide objective guidance for the future. 4.0.2 Variability of tropical cyclone activity/intensity on intraseasonal and interannual scale Topic 4.1 reviews the current status of the understanding on variability of tropical cyclone activity/intensity on intraseasonal to interannual time scales around the globe. Annually, approximately 80−90 tropical cyclones occur over the tropical oceans. The tropical cyclone activities depend on thermodynamic parameters (e.g., sea surface temperature (SST), atmospheric stability, and mid-tropospheric moisture) and dynamic parameters (e.g., low-level vorticity and vertical wind shear). In many cases, thermodynamic parameters are closely linked with each other in the tropics; the atmosphere overlying high SSTs tends to be humid, and humid air with high atmospheric temperature inevitably becomes unstable. Over the tropical oceans prone to frequent tropical cyclones, the thermodynamic factors for tropical cyclone formation are most often satisfied. Also, the dynamic parameters—positive low-level vorticity and weak vertical wind shear—give rise to environments favorable for the generation of tropical cyclones. In the case of changes in the large-scale circulation in the tropical oceans, the thermodynamic and/or dynamic parameters may be modified. These modifications, in turn, may alter the tropical cyclone activity/intensity. The variation of the tropical cyclone activity is to some extent associated with the El Niño–Southern Oscillation, quasi-biennial oscillation, Arctic Oscillation, North Atlantic Oscillation, Antarctic Oscillation, Madden–Julian Oscillation, etc., depending on ocean basins. Subsequently, the discussion will be involve several ocean basins such as the western North Pacific, the North Atlantic, the eastern and central North Pacific, the Indian Ocean, and Australia and South Pacific. 4.0.3 Possible relationships between climate change and tropical cyclone activity Topic 4.2 reviews the current science on possible relationships between climate change and tropical cyclone activity/intensity on different time scales. The report first discusses observed tropical climate trends and multi-decadal variability that are relevant to tropical cyclone activity, including in SSTs, water vapor, atmospheric stability, and atmospheric circulation. Investigations into observed trends and low-frequency variability of tropical cyclone activity are reviewed, which cover all tropical cyclone basins but most focus upon the Atlantic tropical cyclone records. An overview is provided of paleotempestology – the study of pre-historic tropical cyclones using geological proxy evidence or historic documents – which provides localized estimates of tropical cyclone variations on the order of century and millennium timescales. Theoretical studies and numerical models (both coupled global climate models and downscaled regional models) have been utilized extensively to research past tropical cyclone behavior (climatology, seasonal cycles, interannual, decadal and multidecadal variability). These tools have also been instrumental in making projections of future greenhouse gas warming impacts (into tropical cyclone frequency, intensity, and rainfall) as well as allowing assessments of how these changes compare with studies of recent tropical

444

cyclone observations. Finally, the role that tropical cyclones may have in actively forcing the climate system is discussed. There are substantial roadblocks both in making reliable future projections about tropical cyclone activity and in determining whether a trend can be detected in historical tropical cyclone data. For the climate change detection problem, a large hurdle is the quality of the tropical cyclone historical databases. Several recent studies report strong increasing trends in several tropical cyclone metrics. However, the databases used in these studies were unfortunately populated over time without a focus on maintaining data homogeneity, a key requirement for databases which are to be used to assess possible climate-related trends. Additionally, improved understanding of the causes of past variations or trends in tropical cyclone activity will depend on the existence of reliable climate-quality data sets for related variables, such as SST, atmospheric temperature, moisture, wind shear, etc. However, data quality and data inhomogeneity issues with datasets such as the NCAR/NCEP and ERA reanalyses remain as an important roadblock for further advancement. Climate (global and regional) models are another important tool for investigating tropical cyclone climate variability and change. These contain hypotheses for how the climate system works in a framework which allows experiments to be performed to test various hypotheses or compare the model’s historical simulations against historical observations. Nonetheless, there are important uncertainties in climate models and the radiative forcings used for such experiments. Finally, in contrast to the theory of potential intensity of tropical cyclones, which is more well-established, a comparable theory of tropical cyclone frequency is not well-developed at this time. The lack of theoretical underpinning of tropical cyclone genesis and frequency of occurrence remains as an important roadblock to progress in this area, apart from global model limitations. In general, hurricane-climate research is expected to progress most rapidly when a combination of theory, modeling, and observations are brought to bear on the problem. The need for improved climate-quality tropical cyclone databases seems clear. These will provide better information for assessing future changes, and more reliable statistical assessments of past changes in hurricane activity, including landfall, in all basins. Tropical cyclone/climate modeling studies will benefit from efforts to improve global climate modeling in general. In addition, studies which focus on simulation or downscaling of tropical cyclones could benefit from more rigorous testing of model performance with simulating a wider range of tropical cyclone metrics. Finally, exploration of empirical approaches, such as seasonal genesis parameters, should be encouraged, including testing/evaluation and improvements aimed at reproducing characteristics of historical tropical cyclone activity in different basins from both observations and climate model simulations. Based on these results, these approaches may be useful for making climate change projections of tropical cyclone activity. 4.0.4 Short-term climate (seasonal and intra-seasonal) prediction of tropical cyclone activity and intensity Topic 4.3 discusses the state of the science in seasonal and sub-seasonal tropical cyclone prediction. Seasonal tropical cyclone forecasts are currently produced using statistical and dynamical methods in various centers and for different regions. Statistical seasonal tropical cyclone prediction was first conducted in the Atlantic basin at Colorado State University using statistical relationships between Atlantic tropical cyclone activity and predictors such as the El Niño – Southern Oscillation (ENSO), the Quasi-Biennial Oscillation (QBO) and Caribbean basin sea level pressures. Statistical forecast techniques have continued to develop since these early forecasts began in the mid-1980s and have spread to several tropical cyclone basins. Some groups are issuing seasonal forecasts up to almost a year in advance of the season. Verifications of some of the seasonal prediction efforts have demonstrated that substantial skill exists, especially at the shortest lead times. Additionally, recent statistical predictions have been attempting to regionalize the forecasts to impacts along specific coastal zone.

445

Two groups are now issuing seasonal forecasts of tropical storm frequency based on dynamical models. The skill of some dynamical models to predict the frequency of tropical storms over the Atlantic can be comparable to the skill of statistical models. Over the other ocean basins, dynamical models can also display some robust skill in predicting the frequency of tropical storms, but they usually perform poorly over the North and South Indian Oceans. The seasonal prediction of the risk of tropical storm landfall still represents a challenge for dynamical models, as track produced tend to be unrealistically poleward in most modeling systems Interest in the prediction of atmospheric variability on the intra-seasonal timescale has recently blossomed. On this timescale, the Madden-Julian Oscillation (MJO), with its 30- to 80-day period, provides the greatest prospects for tropical prediction given its large scale, tendency to persist for at least an additional cycle, and its moderate to strong relationship to tropical cyclone activity in some basins. MJO prediction has so far been approached using mainly empirical methods, owing to the difficulty that global numerical models have in its simulation and prediction. While there is much room for improvement in the skill and application of empirical/statistical methods of intra-seasonal tropical cyclone prediction, the greatest hope for improvement lies with dynamical/numerical models. Indeed, numerical studies using twin-experiment methodology in which the model employed is assumed to be perfect, indicate useful predictability of the MJO may extend to 25-30 days, 10 days longer than that currently derived from empirical methods.

446

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES

Topic 4.0 : The Recent Active Atlantic Hurricane Seasons Rappateur: Mr. Max Mayfield, Director NOAA/NWS/NCEP/TPC/National Hurricane Center 11691 SW 17th Street Miami, Florida 33165-2149, USA Email: Max.Mayfield@noaa.gov Phone: 1-305-229-4402 Fax: 1-305-553-1901 1. Overview of the 2004’s Atlantic hurricane Season The 2004 Atlantic hurricane season was among the most devastating on record. The year’s storms claimed over 3100 lives, the second largest toll (up through 2004) in three decades; 60 of these occurred in the United States. The United States suffered a record $45 billion in property damage, enduring landfalls from five hurricanes (Charley, Frances, Gaston, Ivan, and Jeanne) and the eyewall passage of a sixth (Alex) that avoided landfall on the North Carolina Outer Banks by less than 10 miles. In addition, Bonnie, Hermine, and Matthew made landfall in the United States as tropical storms. Florida, the “Sunshine State,” became known as the Plywood State after being battered by Charley, Frances, Ivan, and Jeanne. The islands of the Caribbean were also hard hit. Charley struck Cuba as a major hurricane [maximum 1-min winds of greater than 96 kt (1 kt = 0.5144 m s−1), corresponding to category 3 or higher on the Saffir–Simpson hurricane scale (Simpson 1974)]. Ivan was also a major hurricane in the Caribbean, causing extensive destruction on Grenada, Jamaica, and Grand Cayman, and Jeanne produced catastrophic flash floods in Haiti that killed thousands and left hundreds of thousands homeless. Fifteen named storms developed in 2004, including Nicole, a subtropical storm (Table 4.0.1; Figure 4.0.1). Nine of the named systems became hurricanes, and of these, six became major hurricanes. One additional tropical depression did not reach storm strength. These totals are considerably above the long-term (1944–2003) means of 10.2 named storms, 6.0 hurricanes, and 2.6 major hurricanes. August alone saw the formation of eight tropical storms, a new record for that month. The season also featured intense and long-lived hurricanes. Ivan, a category 5 storm, twice reached a minimum pressure of 910 mb, a value surpassed by only five other previous tropical cyclones in the Atlantic basin historical record. In addition, Ivan was a major hurricane for a total of 10 days, a new record for a single storm since reliable records began in 1944. In terms of “accumulated cyclone energy” (ACE; the sum of the squares of the maximum wind speed at 6-h intervals for tropical storms and hurricanes), overall activity this year was 234% of the long-term (1944–2003) mean. Only two seasons since 1944 (1950 and 1995) have had higher ACE values. The 2-month period of August–September 2004 registered the highest 2-month ACE accumulation on record. The above-normal levels of activity in 2004 continued a tendency that began in 1995 for greater numbers of storms. This appears to be due, in part, to sea surface temperatures (SSTs) over the North Atlantic Ocean that have been considerably warmer during the past 10 yr than during the preceding decade. In fact, 2004 was the second warmest year since 1948, as measured by SSTs between 10° and 20°N in the tropical Atlantic Ocean and Caribbean Sea during the peak months of the hurricane season. Nearly the entire tropical Atlantic during the peak of the hurricane season was warmer than normal, the exception being cool anomalies in the extreme western Atlantic that largely reflect up-welling from Frances and Jeanne. Particularly large anomalies were present in the eastern Atlantic

447

north of 15°N; these may have contributed to an unusually favorable thermodynamic environment for tropical waves. Large-scale steering patterns in 2004, however, differed significantly from those occurring over much of the past decade, which had been characterized by a midlevel trough near the eastern coast of the United States that took many storms out to sea before they could make landfall. In contrast, persistent high pressure over the eastern United States and the western Atlantic during 2004 kept this year’s storms on more westerly tracks. The season also featured lower than normal vertical wind shear over the Caribbean Sea and western Atlantic; this combination allowed hurricanes approaching the Caribbean and North America to maintain much of their intensity. 2. Noteworthy 2004 Atlantic Hurricanes Hurricane Charley: 9–14 August - Hurricane Charley strengthened rapidly just before striking the southwestern coast of Florida as a category 4 hurricane. Charley was the strongest hurricane to hit the United States since Andrew in 1992 and, although small in size, caused catastrophic wind damage in Charlotte County, Florida. Serious damage occurred well inland over the Florida peninsula. Hurricane Frances: 25 August–8 September - Frances was a Cape Verde tropical cyclone that passed through the Bahamas as a major hurricane and struck the Florida east coast as a category 2 hurricane. Hurricane Ivan: 2–24 September - Ivan, one of the strongest tropical cyclones on record in the Atlantic basin, was a long-lived Cape Verde hurricane that reached category 5 strength on three occasions. Ivan carved a path of destruction through the Caribbean, striking Grenada, Jamaica, the Cayman Islands, and Cuba as a major hurricane. Ivan also made landfall as a major hurricane in the United States, causing over $14 billion in damage. Hurricane Jeanne: 13–28 September - Jeanne produced catastrophic flash floods in Haiti that killed over 3000 people, and later struck the east coast of Florida as a major hurricane. Jeanne was the fourth hurricane to hit the state of Florida in 2004, and the second to strike Florida’s Treasure Coast in a 3-week span. 3. Overview of 2005’s Atlantic hurricane Season By almost all standards of measure, the 2005 Atlantic hurricane season was the most active of record. Twenty-eight storms – twenty-seven tropical and one subtropical – formed during the year (Table 4.0.2 and Figure 4.0.2). This broke the record of 21 set in 1933. Fifteen of the storms became hurricanes, breaking the record of twelve set in 1969. Seven hurricanes became major hurricanes corresponding to category 3 or higher on the Saffir-Simpson hurricane scale (Simpson 1974). This was just short of the record of eight major hurricanes set in 1950 (Norton 1951). Four hurricanes reached category 5 strength [maximum 1-min winds greater than 135 kt], which was the first time this had been observed in one season. In terms of “accumulated cyclone energy” (ACE: the sum of the squares of the maximum wind speed at 6-h intervals for named storms and hurricanes), overall activity this year was the highest of record, about 256% of the long-term (1944-2003) mean. The previous record was about 249% of the long term mean set in 1950. A record seven named storms formed before the end of July, including Hurricane Emily, the earliest category 5 hurricane on record in the basin. The season also ran late, as Tropical Storm Zeta was the second latest developing storm of record and lasted into January 2006. Three of the category 5 hurricanes featured very low (by Atlantic basin standards) minimum central pressures. The central pressure of Hurricane Wilma was estimated from aircraft data to be 882 mb. This is the lowest known central pressure in an Atlantic hurricane, eclipsing the pressure of 888 mb observed in Hurricane Gilbert in 1988 (Lawrence and Gross 1989). Additionally, Hurricane Rita had an estimated minimum central pressure of 895 mb (the fourth lowest of record), while Hurricane Katrina had a minimum central pressure of 902 mb (the sixth lowest of record). As might be expected, these

448

storms all underwent periods of rapid to explosive deepening. The central pressure of Wilma fell 88 mb in about 12 h, shattering most records for deepening rates of Atlantic tropical cyclones. The central pressure of Rita fell 70 mb in about 24 h, while the central pressure of Katrina fell 46 mb in about 24 h. On a shorter time-scale, the central pressure of Hurricane Dennis fell 11 mb in 1h 35 min. While a full explanation of the record level of activity will require considerable research, there are several likely contributing factors. One was the warmth of the North Atlantic Ocean, as sea-surface temperatures (SSTs) during the hurricane season were the warmest ever observed in the Caribbean Sea and tropical Atlantic Ocean. Much of the tropical Atlantic had above normal SSTs, with anomalies of up to 1oC in the western part of that area. Above normal anomalies were also present through the Caribbean Sea and the Gulf of Mexico. It should be noted that the lower-magnitude warm anomalies in the eastern Gulf of Mexico are likely due to cooling caused by repeated hurricane passages. A striking contrast between the 2004 and 2005 hurricane seasons was the amount of activity in the tropical Atlantic. During the 2004 season (Franklin et al. 2006), this area saw the formation of three long-track major hurricanes (Frances, Ivan, and Karl) and several less intense systems. Despite the great activity during 2005, only two hurricanes (Emily and Philippe - both category 1) occurred in this region. This was not due to lack of chances, as five named storms and two depressions formed in the region, and two other named storms began over the eastern Caribbean Sea. A possible factor in the large number of landfalling cyclones during 2005 - particularly the record four major hurricanes to hit the United States - is stronger-than-average ridging in the middle troposphere that persisted over the eastern United States during in a fashion somewhat similar to that seen in 2004. This ridge was a little south and west of that seen in 2004 and likely helped steer hurricanes farther west and south into the Gulf of Mexico during 2005. The 2005 hurricane season was the deadliest in the Atlantic basin since 1998. Katrina is believed to be directly responsible for 1500 deaths, while heavy rains associated with a large area of disturbed weather during Hurricane Stan produced flooding that cause 1000-2000 deaths. The season caused over $100 billion dollars in property damage in the United States alone, making it the costliest season of record. 4. Noteworthy 2005 Atlantic Hurricanes Hurricane Dennis: 4-13 July - Hurricane Dennis was an unusually strong July major hurricane that left a trail of destruction from the Caribbean Sea to the northern coast of the Gulf of Mexico. Hurricane Emily: 11-21 July - Emily was briefly a category 5 hurricane in the Caribbean Sea that, at lesser intensities, struck Grenada, resort communities on Cozumel and the Yucatan Peninsula, and northeastern Mexico just south of the Texas border. Emily is the earliest-forming category 5 hurricane on record in the Atlantic basin and the only known hurricane of that strength to occur during the month of July. Hurricane Katrina: 23-30 August - Katrina was an extraordinarily powerful hurricane that carved a wide swath of catastrophic damage and inflicted large loss of life. It was the costliest and one of the five deadliest hurricanes to strike the United States. Katrina first caused fatalities and damage in southern Florida as a category 1 hurricane. After reaching category 5 intensity over the central Gulf of Mexico, Katrina weakened to category 3 before making landfall on the northern Gulf coast. Even so, the damage and loss of life inflicted by this massive hurricane in Louisiana and Mississippi were staggering. Also, significant effects extended into the Florida Panhandle, Georgia, and Alabama. Considering the scope of its impacts, Katrina was one of the most devastating natural disasters in United States history.

449

Hurricane Rita: 18-26 September - Rita was an intense hurricane that reached category 5 strength over the central Gulf of Mexico, where it had the fourth-lowest central pressure on record in the Atlantic basin. Although it weakened before making landfall as a category 3 hurricane near the Texas/Louisiana border, Rita produced significant storm surge that devastated coastal communities in southwestern Louisiana, and its winds, rain, and tornadoes caused fatalities and a wide swath of damage from eastern Texas to Alabama. Additionally, Rita caused storm surge flooding in portions of the Florida Keys. Hurricane Stan: 1-5 October - Stan was associated with disastrous inland flooding across portions of Central America and Mexico, and some estimates of the death toll are as high as 2000. However, not all of these deaths can be directly attributed to Stan. Hurricane Wilma: 15-25 October - Wilma formed and became an extremely intense hurricane over the northwestern Caribbean Sea. It had the all-time lowest central pressure for an Atlantic basin hurricane, and it devastated the northeastern Yucatan Peninsula. Wilma also inflicted extensive damage over southern Florida. 5. Monitoring of Atlantic Tropical Cyclones in 2004 and 2005 For storms east of 55oW longitude, or those not threatening land, the primary (and often sole) source of information is geostationary and polar-orbiting weather satellite imagery, interpreted using the Dvorak (1984) and Hebert-Poteat (1975) intensity estimation techniques. These estimates (“classifications”) are provided by the Tropical Analysis and Forecast Branch (TAFB) of the Tropical Prediction Center (TPC), the Satellite Analysis Branch (SAB) in Washington, DC, and the Air Force Weather Agency (AFWA) in Omaha, NE. For systems threatening land, in situ observations are also generally available from aircraft reconnaissance flights conducted by the 53rd Weather Reconnaissance Squadron (WRS, “Hurricane Hunters”) of the Air Force Reserve Command (AFRC), and by the National Oceanic and Atmospheric Administration (NOAA) Aircraft Operations Center (AOC). During reconnaissance flights, minimum sea-level pressures are either measured by dropsondes released at the circulation center or extrapolated hydrostatically from flight-level. Surface (or very near-surface) winds in the eyewall or maximum wind band can be measured directly using Global Positioning System (GPS) dropwindsondes (Hock and Franklin 1999), but more frequently are estimated from flight-level winds using empirical relationships derived from a three-year sample of GPS dropwindsonde data (Franklin et al. 2003). During NOAA reconnaissance missions, surface winds can be estimated remotely using the Stepped-Frequency Microwave Radiometer (SFMR) instrument (Uhlhorn and Black 2003). When available, satellite and reconnaissance data are supplemented by conventional land-based surface and upper-air observations, ship and buoy reports, and weather radars, including the U. S. National Weather Service’s (NWSs) network of Doppler radars. In key forecast situations, the kinematic and thermodynamic structure of the storm environment is obtained from dropwindsondes released during operational “synoptic surveillance” flights of NOAA’s Gulfstream IV jet aircraft (Aberson and Franklin 1999). Several satellite-based remote sensors play an important role in the analysis of tropical weather systems. Foremost of these is multi-channel passive microwave imagery [e.g., from the Tropical Rainfall Measuring Mission (TRMM) satellite], which over the past decade has provided radar-like depictions of systems’ convective structure (Hawkins et al. 2001), and is of great help in assessing system location and organization. The SeaWinds scatterometer onboard the QuikSCAT satellite (Tsai et al. 2000) provides surface winds over large oceanic swaths. While the QuikSCAT generally does not have the horizontal resolution to determine a hurricane’s maximum winds, it can sometimes be used to estimate the intensity of weaker systems and to determine the extent of tropical storm force winds. In addition, it can be helpful in determining whether an incipient tropical cyclone has acquired a closed surface circulation. Finally, information on the thermal structure of cyclone cores is provided by the Advanced Microwave Sounder Unit (AMSU, Velden and Brueske 1999). Intensity estimates

450

derived from such data can in some cases be superior to Dvorak classifications (Herndon and Velden 2004). A number of organizations have developed web sites that are helpful for tropical cyclone forecasting and post-analysis. These include the NRL Monterey Marine Meteorology Division Tropical Cyclone Page (http://www.nrlmry.navy.mil/tc_pages/tc_home.html), with its comprehensive suite of microwave products, the cyclone phase diagnostics page of Florida State University (http://moe.met.fsu.edu/cyclonephase/), which is frequently consulted to help categorize systems as tropical, subtropical, or non-tropical, and the tropical cyclone page of the University of Wisconsin/Cooperative Institute for Meteorological Satellite Studies (http://cimss.ssec.wisc.edu/tropic/tropic.html), which contains a variety of useful satellite-based synoptic analyses. In the cyclone summaries, U. S. damage estimates have been generally estimated by doubling the insured losses reported by the American Insurances Service Group (AISG) for events exceeding their minimum reporting threshold ($25 million). However, the uncertainty in estimating meteorological parameters for tropical cyclones pales in comparison to the uncertainty in determining the cost of the damage they cause. Descriptions of the type and scope of damage are taken from a variety of sources, including local government officials, media reports, and local NWS Weather Forecast Offices (WFOs) in the affected areas. Official NHC Tropical Cyclone Reports can be obtained at: http://www.nhc.noaa.gov/pastall.shtml . REFERENCES Aberson, S. D., and J. L. Franklin, 1999: Impact on hurricane track and intensity forecasts of GPS dropwindsonde observations from the first season flights of the NOAA Gulfstream-IV jet aircraft. Bull. Amer. Meteor. Soc., 80, 421-427. Dvorak, V. E., 1984: Tropical cyclone intensity analysis using satellite data. NOAA Tech. Rep. NESDIS 11, National Oceanic and Atmospheric Administration, Washington, DC, 47 pp. Franklin, J. L., M. L. Black and K. Valde, 2003: GPS dropwindsonde wind profiles in hurricanes and their operational implications. Wea. Forecasting 18, 32-44. --------, R. J. Pasch, L. A. Avila, J. L. Beven II, M. B. Lawrence, S. R. Stewart, and E. S. Blake, 2006: Atlantic hurricane season of 2004. Mon. Wea. Rev., 134, 981-1025. Hawkins, J. D., T. F. Lee, J. Turk, C. Sampson, F., J. Kent, and K. Richardson 2001: Real-time internet distribution of satellite products for tropical cyclone reconnaissance. Bull. Amer. Meteor. Soc., 82, 567-578. Hebert, P. J., and K. O. Poteat, 1975: A satellite classification technique for subtropical cyclones. NOAA Technical Memorandum NWS SR-83, U.S. Dept. of Commerce, National Weather Service, Ft. Worth TX, 25 pp. Herndon, D. C., and C. Velden, 2004: Upgrades to the UW-CIMSS AMSU-based tropical cyclone intensity estimation algorithm. Preprints, 26th Conf. Hurr. Trop. Meteor., Miami, Amer. Meteor. Soc., 118-119. Hock, T. F., and J. L. Franklin, 1999: The NCAR GPS dropwindsonde. Bull. Amer. Meteor. Soc., 80, 407-420.

451

Lawrence, M. B., and J. M. Gross: Atlantic Hurricane Season of 1988. Mon. Wea. Rev., 117, 2248-2259. Norton, G., 1951: Hurricanes of the 1950 season. Mon. Wea. Rev., 79, 8-15. Simpson, R. H., 1974: The hurricane disaster potential scale. Weatherwise, 27, 169 & 186. Tsai, W.-Y., M. Spender, C. Wu, C. Winn and K. Kellogg, 2000: SeaWinds of QuikSCAT: Sensor description and mission overview. Proceedings, Geoscience and Remote Sensing Symposium 2000, IGARSS 2000, IEEE 2000 International, Vol 3., 1021-1023. Uhlhorn, E. W., and P. G. Black, 2003: Verification of remotely sensed sea surface winds in hurricanes. J. Atmos. and Ocean. Tech., 20, 99-116. Velden, C. S., and K. F. Brueske, 1999: Tropical cyclone warm cores as observed from the NOAA polar orbiting satellite’s new Advanced Microwave Sounder Unit. Preprints, 23rd Conf. Hurr. Trop. Meteor., Dallas, Amer. Meteor. Soc., 182-185.

452

TABLE 4.0.1. 2004 Atlantic hurricane season statistics.

453

Table 4.0.2. Atlantic Hurricanes, Tropical Storms, and Subtropical Storms of 2005.

Name Classa Datesb

Maximum 1-min wind (kt)

Minimum sea level pressure (mb)

Direct deaths

U. S. damages ($ million)

Arlene T 8 - 13 Jun 60 989 1 Minorc Bret T 28 - 30 Jun 35 1002 1 Cindy H 3 - 7 Jul 65 991 1 320 Dennis H 4 - 13 Jul 130 930 42 2,230 Emily H 11 - 21 Jul 140 929 6 Minorc Franklin T 21 - 29 Jul 60 997 Gert T 23 - 25 Jul 40 1005 Harvey T 2 - 8 Aug 55 994 Irene H 4 - 18 Aug 90 970 Jose T 22 - 23 Aug 50 998 6 Katrina H 23 - 30 Aug 150 902 1500 81,000 Lee T 28 Aug- 2 Sep 35 1006 Maria H 1 - 10 Sep 100 962 Nate H 5 - 10 Sep 80 979 Ophelia H 6 - 17 Sep 75 976 1 70 Philippe H 17 - 24 Sep 70 985 Rita H 18 - 26 Sep 155 895 7 10,000 Stan H 1 - 5 Oct 70 977 80 Unnamed ST 4 - 5 Oct 45 997 Tammy T 5 - 6 Oct 45 1001 Minorc Vince H 8 - 11 Oct 65 988 Wilma H 15 - 25 Oct 160 882 23 20,600 Alpha T 22 - 24 Oct 45 998 26 Beta H 26 - 31 Oct 100 962 Gamma T 14 - 21 Nov 45 1002 37 Delta T 22 - 28 Nov 60 980 Epsilon H 29 Nov-8 Dec 75 981 Zeta T 30 Dec -6 Jan 55 994

a T = tropical storm and ST = subtropical storm, wind speed 34-63 kt (17-32 m s-1); H = hurricane, wind speed 64 kt (33 m s-1) or higher. b Dates begin at 0000 UTC and include tropical and subtropical depression stages but exclude extratropical stage. c Only minor damage was reported, but the extent of the damage was not quantified.

454

Figure 4.0.1. Atlantic Hurricanes, Tropical Storms, and Subtropical Storms of 2004.

455

Figure 4.0.2. Atlantic Hurricanes, Tropical Storms, and Subtropical Storms of 2005.

456

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES Topic 4.1 : Variability of Tropical Cyclone Activity/Intensity on Intraseasonal and Interannual Scales Rapporteur: Chang-Hoi Ho Seoul National University, Republic of Korea E-mail : hoch@cpl.snu.ac.kr Working Group: J H Kim, C H Sui, Gerry Bell, Jeff Callaghan Topic 4.1 reviews the current status of the understanding on variability of tropical cyclone (TC) activity/intensity on intraseasonal to interannual time scales around the globe. Annually, approximately 80−90 TCs occur over the tropical oceans (Neumann 1993). The TC activities depend on thermodynamic parameters (e.g., sea surface temperature (SST), atmospheric stability, and mid-tropospheric moisture) and dynamic parameters (e.g., low-level vorticity, vertical wind shear, and upper-tropospheric momentum flux convergence) (Gray 1979). In many cases, thermodynamic parameters are closely linked with each other in the tropics; the atmosphere overlying high SSTs tends to be humid, and humid air with high atmospheric temperature inevitably becomes unstable. Over the tropical oceans prone to frequent TCs, the thermodynamic factors for TC formation are most often satisfied. Also, the dynamic parameters—positive low-level vorticity and weak vertical wind shear—give rise to environments favorable for the generation of TCs.. In the case of changes in the large-scale circulation in the tropical oceans, the thermodynamic and/or dynamic parameters may be modified. These modifications, in turn, may alter the TC activity/intensity. The variation of the TC activity is to some extent associated with the El Nino–Southern Oscillation (ENSO), quasi-biennial oscillation (QBO), Arctic Oscillation (AO), North Atlantic Oscillation (NAO), Antarctic Oscillation (AAO), Madden–Julian oscillation (MJO), etc., depending on ocean basins. Subsequently, the discussion will be involve several ocean basins such as the western North Pacific (WNP), the North Atlantic (NA), the eastern and central North Pacific (ECNP), the Indian Ocean (IO), and Australia and South Pacific (ASP). a) Western North Pacific (WNP) A pronounced interannual variability exists in the TC activity in the WNP in connection to the ENSO events; the active genesis region of the TC moves both eastward and toward the equator, the life span and probability of the intense TC increase, and TCs recurve more often and tend to recurve farther eastward during the warm phase of ENSO (e.g., Chan 1985; Lander 1994; Chen et al. 1998; Chan 2000; Chia and Ropelewski 2002; Wang and Chan 2002; Camargo and Sobel 2005). The increase (decrease) in the TC activity in the southeast (northwest) quadrant of the WNP during El Nino years can be explained by the eastward shift in the preferred genesis location and the increased northward steering flows. The opposite is true during La Nina years. There is also a notable frequency reduction in TC formation in the summer following of the El Nino year, corresponding to a longitudinal shift of the Walker circulation (Chan 1985; Wu and Lau 1992; Chan 2000). The cross-spectral analysis demonstrated that two time series of the stratospheric QBO and TC activity in the WNP are almost in phase during the 28-month period (Chan 1995); the westerly phase of the QBO corresponds to a larger number of TCs. The QBO−TC relation would be a result of the decrease

457

in the upper-tropospheric vertical shear over the tropics during the boreal summer associated with the westerly phase of the QBO. However, this relationship did not hold true during ENSO years. Ho et al. (2005) suggested that the TC activity in the WNP is also influenced by the large-scale circulation in the Southern Hemisphere (e.g., AAO) through changes in convection activity along the Intertropical Convergence Zone in the western equatorial Pacific. In the positive phase of AAO (relative to its negative phase), two anomalous highs— a huge anticyclone in southeastern Australia and a relatively weak anticyclone in the East China Sea—develop over the western midlatitude Pacific in both the hemispheres. A statistically significant alteration in TC activity related to the AAO variations is observed over the WNP. This change has primarily resulted due to the increase in TC formation over the eastern Philippine Sea. The number of TCs in the WNP and TC landfalls in China is seen to negatively correlate with the Tibetan Plateau snow cover during the preceding winter and spring (Xie et al. 2005). When the Tibetan Plateau snow cover exceeded normal levels, fewer TCs were formed in the WNP and made landfall in China and vice versa. This can be understood by the response of the western Pacific subtropical high to the snow-modulated land surface thermodynamic processes over the Tibetan Plateau. On an intraseasonal time scale, the period of active TC genesis tends to form a cluster over a period of 2−3 weeks, followed by a comparable period of inactive TC genesis (Gray 1979). Hence, by combining both the active and inactive periods, the TC activity seems to fluctuate over a 4−6 week period, i.e., the periodicity on intraseasonal time scale. Over the WNP, the TC activity tends to be strong during the MJO convective period (e.g., Nakazawa 1986; Hartmann et al. 1992; Liebmann et al. 1994). However, the ratio of intense systems (i.e., tropical storms and typhoons) to weak systems (i.e., tropical depressions) seems nearly constant regardless of the convective and/or dry periods (Liebmann et al. 1994). The enhanced TC activity during the MJO convective period can be explained by low-level barotropic wave dynamics over monsoon confluent regions (e.g., Holland 1995; Sobel and Maloney 2000; Maloney and Hartmann 2001; Maloney and Dickinson 2003). The time-mean values of the barotropic wave-activity flux at 850 hPa increases during the westerly phase of the MJO. When an anomalous westerly phase occurs in the lower tropospheric circulation field, a barotropic convection process from the mean flow to the eddy kinetic energy is activated; this process is associated with the strong low-level convergence of the monsoon flow. Thus, eddies grow such that a favorable condition for TC genesis is provided. The opposite is true during an anomalous easterly phase. Further, Kim et al. (2006) found that TC tracks largely depend on the MJO phases; when the MJO-related convection center lacates in the equatorial Indian Ocean (tropical WNP), TC passages migrate eastward (westward) due to changes in both the major genesis region and prevailing large-scale steering flows. There are also efforts to understand TCs variability on an intraseasonal time scale by classifying their tracks in connection with large-scale circulation. Hodanish and Gray (1993) stratified four patterns according to the difference in the recurving process—sharply recurving cyclones, gradually recurving cyclones, left-turning cyclones, and nonrecurving cyclones. Harr and Elsberry (1991, 1995a, b) elucidated that certain pattern of TC tracks can be distinguished based on the anomalous large-scale circulation. Their patterns were classified into three classes— straight, recurving south (recurving TCs that formed south of 20°N), and recurving north (recurving TCs that formed north of 20°N). Lander (1996) also considered categorizations of TC motion which he classified into four major patterns—straight moving, recurving, north-oriented, and staying—in the South China Sea. Kim et al. (2005) investigated the variation of the summertime TC activity over East Asia using the empirical orthogonal function analysis and found a pronounced west-east oscillation between Korea and Japan. Anomalous atmospheric flows connected to the west-east oscillation are an enhanced anticyclonic (cyclonic) circulation centered on Japan when the TC activity is high over the south of Korea (southeast of Japan), showing an equivalent barotropic structure in the whole troposphere.

458

b) North Atlantic (NA) In many cases, TCs in the NA are developed associated with African easterly waves (e.g., Landsea and Gray 1992; Goldenberg and Shapiro 1996; Thorncroft and Hodges 2001). The activity of African easterly waves intensifies (weaker) when the rainfall amount over the western Sahel region exceeds (below) normal. Namely, TC activity in the NA seems to be increased during wet years of the western Sahel. Based on the statistics of the tracking of vorticity centers connected to African easterly wave activity, Thorncroft and Hodges (2001) showed that the 850-hPa easterly wave at the West African coast between about 10°N and 15°N is highly correlated to TC activity in the NA. This correlation is particularly strong for the period 1994−1998. This indicates that Atlantic tropical cyclone activity is not influenced only by the total number of African easterly waves but also by the number of African easterly waves that leave the West African coast, which have significant low-level amplitudes. The western Sahel rainfall is well correlated with ENSO events. Hence, most parts of the interannual variation in the TC activity in the NA would also be understood by an eastward shift of warm SST regions to the eastern Pacific and corresponding changes in the large-scale convection associated with ENSO (e.g., Shapiro 1987; Goldenberg and Shapiro 1996; Tang and Neelin 2004). The ENSO−TC relation results from changes in the vertical wind shear—an enhanced divergent outflow from deep cumulus convection during El Nino years results in an increase in westerly wind in the upper troposphere over the Caribbean and tropical Atlantic; however, variations in the lower tropospheric easterly winds are relatively small. Combining these different influences on zonal wind in the upper and lower troposphere, the vertical wind shear over the NA increases during El Nino years as compared to La Nina years (Gray and Sheaffer 1991); consequently, the number of TCs and their duration are reduced during El Nino (Landsea et al. 1999). Tang and Neelin (2004) suggested that the anomalous tropospheric temperatures arising communicated the Pacific due to wave dynamics influence the TC development by affecting column stability relative to equilibrium with NA SST. Bell and Chelliah (2006) further investigated into the tropospheric circulation and SST changes that are linked to Atlantic tropical cyclones, both on interannual and interdecadal timescales. Larson et al. (2006) indicated that the AO (and/or NAO) has also acquired a strong influence on the interannual and intraseasonal variability of TC activity in the NA—an enhanced (decreased) TC activity during the positive (negative) phase of the AO. In the positive phase of the AO, the subtropical ridge in the NA is enhanced; a weakening of the Hudson Bay low in the eastern United States and a strengthening and westward extension of the Bermuda high in the western NA. The westerly wind shear is weakens over the main developing region and the tropical easterly jet intensifies over Africa. All of these characteristics provide favorable conditions for TC development. Interestingly, large-scale circulations for the positive (negative) phase of the AO appear to be similar to those for La Nina (El Nino). Namely, during La Nina years, large-scale circulation is more conductive to TC development during the AO-positive phase than during the negative phase and, during El Nino years. Therefore, it is less conductive to the TC development during the AO-negative phase than during the positive phase. The influence of the QBO on the TC activity is known to be pronounced in the NA than in the other ocean basins (e.g., Gray 1984; Gray et al. 1992; Elsner et al. 1999). During the westerly (easterly) phase of the QBO, the strong TC genesis (i.e., hurricanes) frequency is above (below) normal. It is hypothesized that the ventilation processes in the horizontal wind across the top of the TC are a possible physical mechanism of the QBO-related change. The speed of the zonal wind in the tropical stratosphere is weak during the westerly phase of the QBO. In this case, there is relatively less ventilation resulting in a positive effect on the TC development. In addition, Shapiro (1989) demonstrated that the largest correlations between storm activity in the NA and the 30 hPa wind are observed in June. This indicates that the TCs tend to attain a higher intensity when the QBO is in its westerly phase in the tropical lower stratosphere. Recently, however, it is noted that at least in the Atlantic, the QBO is no longer being utilized for

459

seasonal hurricane forecasting.. The QBO−Atlantic hurricane relationship that Gray (1984) identified from 1950 to 1983 disappeared from 1984 to present date. Thus, NOAA does not consider the QBO phase for TC forecasting (Landsea, C., personal communication) Maloney and Hartmann (2000a) suggested that the MJO is the strongest influencing factor on the intraseasonal variation of TC activity in the NA. During the westerly phase of the MJO, strong anomalous westerlies are observed in the eastern Pacific extending to the western Caribbean being altered by the southwesterlies, resulting in cyclonic circulation anomalies over the Gulf of Mexico. In this period, greatly enhanced TC activity is observed over the Gulf of Mexico and western Caribbean due to the increased genesis frequency of TC over these regions. During the easterly phase of the MJO, the TC genesis is significantly suppressed because anticyclonic circulation anomalies are formed over those regions. c) Eastern and Central North Pacific (ECNP) The TC activity in the ECNP is also related to the phase of ENSO and the MJO. During El Nino years, as Irwin and Davis (1999) noted, changes in the warm SST regions lead to the westward shift in the genesis location of TCs in the eastern North Pacific, thereby resulting in the propagation of TCs farther west into the central Pacific. Concurrently, there is a significant modification in the vertical shear; the decrease in the vertical shear over the tropical central North Pacific during El Nino years. The weaker vertical shear provides favorable conditions for TCs and increases the likelihood of TCs to propagate into the central North Pacific. As a result, the TC activity in the central North Pacific exceeds the normal during El Nino years and is below the normal during La Nina years (Wu and Lau 1992). For the non-El Nino years, most TCs follow a westward or northwestward track (Chu and Wang 1997; Chu 2005). The MJO-related modulation in TC activity in the eastern North Pacific is similar to the modulation in the NA (e.g., Molinari and Vollaro 2000; Maloney and Hartmann 2000b, 2001). Over twice as many TCs are formed during the westerly periods of the MJO as compared to the easterly periods, and the TC intensity is also intensifies. Therefore, the number of hurricanes during the westerly periods is over four times larger than those during the easterly periods. d) Indian Ocean (IO) The TC activity in the North Indian Ocean (NIO) is normally dependent upon the monsoon depression activity (Goswami et al. 2003). The MJO is an important modulator, which influences the clustering of the monsoon depression. During the active period of the MJO, the formation frequency of monsoon depression increases due to enhanced low-level cyclonic shear vorticity related to the westerly MJO winds. However, the evolution from the monsoon depression to TC is slightly difficult because the main developing region is adjacent to land. Over the South Indian Ocean (SIO), during warm ENSO periods, the TC genesis was shifted westward, enhancing the TC formation west of 75°E and reducing east of 75°E. These changes in the TC genesis correspond to a westward shift of convection (Ho et al. 2006). This is explained by a remote effect on the SIO; the increase in SST in the central-eastern Pacific alters the Walker circulation and forms an anomalous anticyclonic circulation in the east SIO during El Nino years (Jury 1993). The spatial difference in the TC passages between El Nino and La Nina shows a significant decrease in the southeast of Madagascar but a moderate increase in the central midlatitude SIO, indicating that TCs move further east during El Nino years (Ho et al. 2006). The changes in the TC activity in the SIO may be attributed to the inclusion of weak intensity systems (vmax < 32 ms–1) that are considered as TCs. It is known that ENSO increases SSTs in the warm phase in the central SIO (Lau and Nath 2003). However, destructive TCs of category 3 and higher (vmax > 32 ms−1) may be suppressed via the vertical westerly wind shear and reduced upper anticyclonic vorticity that are generated during El Nino years (Jury 1993).

460

The variation in the TC activity is also found to depend on various MJO phases (Bessafi and Wheeler 2006; Ho et al. 2006): frequent TC passages for phases 2−4 (strong convective activity straddles along the equatorial Indian Ocean) versus infrequent TC passages for additional phases. TC tracks are more south-oriented in phase 3 as compared to those in phases 2 and 4. This is possibly caused by the increased steering northerlies which are a part of the anticyclonic Gill-type Rossby wave in response to the suppressed MJO-related convection in the maritime continent. The frequency of the TC genesis in the western SIO increases during the east phase of the QBO (Jury 1993; Jury et al. 1999). The Walker circulation anomaly with upper-level easterlies and lower westerlies develop in the tropics north of Madagascar in where the SSTs exceed the normal. In the subtropical region, both the trade easterlies and Hadley circulation are enhanced with a poleward shift of the midlatitude westerlies. It is noted that the QBO is in phase with ENSO approximately every 4 years with the QBO leading every 4 months. Thus, the QBO periodically exerts a similar influence on the TC activity in the western SIO. e) Australia and the South Pacific (ASP) The TC activity in the Australian region is higher during La Nina years and below normal average during El Nino years (e.g., Evans and Allan 1992; Nicholls et al. 1998; Kuleshovand de Hoedt 2003). This ENSO−TC relation is obtained from a strong correlation between the sea level pressure in Darwin, Australia, and TC days around the Australian region. The MJO also strongly modulates the TC activity in the ASP with pronounced modulation to the northwest of Australia, i.e., there are significantly more TCs formed during the active phase of the MJO. This relationship is strengthened during El Nino periods (Hall et al. 2001). References Bell, G. D., and M. Chelliah, 2006: Leading tropical modes associated with interannual and multidecadal fluctuations in North Atlantic hurricane activity. J. Climate, 19, 590–612.

Bessafi, M., and M. C. Wheeler, 2006: Modulation of south Indian Ocean tropical cyclone by the Madden-Julian oscillation and convectively-coupled equatorial waves. Mon. Wea. Rev., 134, 638–656.

Camargo, S. J., and A. H. Sobel, 2005: Western North Pacific tropical cyclone intensity and ENSO. J. Climate, 18, 2996–3006.

Chan, J. C. L., 1985: Tropical cyclone activity in the Northwest Pacific in relation to the El Nino/Southern Oscillation phenomenon. Mon. Wea. Rev., 113, 599–606.

Chan, J. C. L., 1995: Tropical cyclone activity in the western North Pacific in relation to the stratospheric quasi-biennial oscillation. Mon. Wea. Rev., 123, 2567–2571.

Chan, J. C. L., 2000: Tropical cyclone activity over the western North Pacific associated with El Niño and La Niña Events. J. Climate, 13, 2960–2972.

Chen, T.-C., S.-P Weng, N. Yamazaki, and S. Kiehne, 1998: Interannual variation in the tropical cyclone formation over the western North Pacific. Mon. Wea. Rev., 126, 1080–1090.

Chia, H.-H., and C. F. Ropelewski, 2002: The interannual variability in the genesis location of tropical cyclones in the northwest Pacific. J. Climate, 15, 2934-2944.

Chu, P. S. and J. Wang, 1997: Tropical cyclone occurrences in the vicinity of Hawaii: Are the

461

differences between El Nino and non-El Nino years significant? J. Climate, 10, 2683–2689.

Chu, P.-S., (2005), ENSO and tropical cyclone activity, in Hurricanes and Typhoons: Past, Present, and Potential, edited by R.J. Murnane and K.B. Liu, pp. 297–332, Columbia University Press, New York.

Elsner, J. B., A. B. Kara, and M. A. Owens, 1999: Fluctuations in North Atlantic hurricane frequency. J. Climate, 12, 427–437.

Evans, J. L., and R. J. Allan, 1992: El Nino/Southern Oscillation modification to the structure of the monsoon and tropical cyclone activity in the Australian region. Intl. J. Climatol., 12, 611–623.

Goldenberg, S. B., and L. J. Shapiro, 1996: Physical mechanisms for the association of El Nino and West Africa rainfall with Atlantic major hurricanes. J. Climate, 9, 1169–1187.

Goswami, B. N., R. S. Ajayamohan, P. K. Xavier, and D. Sengupta, 2003: Clustering of synoptic activity by Indian summer monsoon intraseasonal oscillations. Geophys. Res. Lett., 30, 1431–1434.

Gray, W. M., 1979: Hurricanes: Their formation, structure and likely role in the general circulation. Meteorology Over the Tropical Oceans, edited by D.B. Shaw, Ed., Royal Meteorological Society, James Glaisher House, Grenville Place, Bracknell, Berks, RG 12 1BX, 155–218.

Gray, W. M., 1984: Atlantic seasonal hurricane frequency. Part I: El Nino and 30 mb Quasi-Biennial Oscillation influences. Mon. Wea. Rev., 112, 1649–1168.

Gray, W. M., C. W. Landsea, P. W. Mielke, Jr., and K. J. Berry, 1992: Predicting Atlantic seasonal hurricane activity 6–11 months in advance. Wea. Forecasting, 7, 440–455.

Gray, W. M., and J. D. Sheaffer, 1991: El Nino and QBO influences on tropical cyclone activity. In Teleconnections linking worldwide climate anomalies, edited by M H. Glantz, R. W. Katz, and N. Nicholls, 257–284, Cambridge University Press, New York.

Hall, J., A. J. Matthews, and D. Karoly, 2001: The modulation of tropical cyclone activity in the Australian region by the Madden-Julian oscillation. Mon. Wea. Rev., 129, 2970–2982.

Harr, P. A. and R. L. Elsberry, 1991: Tropical cyclone track characteristics as a function of large-scale circulation anomalies. Mon. Wea. Rev., 119, 1448–1468.

Harr, P. A., and R. L. Elsberry, 1995a: Large-scale circulation variability over the tropical western North Pacific. Part I: Spatial patterns and tropical cyclone characteristics. Mon. Wea. Rev., 123, 1225–1246.

Harr, P. A., and R. L. Elsberry, 1995b: Large-scale circulation variability over the tropical western North Pacific. Part II: Persistence and transition characteristics. Mon. Wea. Rev., 123, 1247–1268.

Hartmann, D. L., M. L. Michelsen, and S. A. Klein, 1992: Seasonal variations of tropical intraseasonal oscillations: A 20−25-day oscillation in the western Pacific. J. Atmos. Sci., 49, 1277–1289.

Ho, C.-H., J.-H. Kim, H.-S. Kim, C.-H. Sui, and D.-Y. Gong, 2005: Possible influence of the Antarctic Oscillation on tropical cyclone activity in the western north Pacific. J. Geophys. Res., 110, D19104, doi:10.1029/2005JD005766.

Ho, C.-H., J.-H. Kim, J.-H. Jeong, H. S. Kim, and D. Chen, 2006: Variation of tropical cyclone activity in the south Indian Ocean: ENSO and MJO effects. J. Geophys. Res., in press.

Hodanish, S., and W. M. Gray, 1993: An observational analysis of tropical cyclone recurvature. Mon. Wea. Rev., 121, 2665–2689.

462

Holland, G. J., 1995: Scale interaction in the western Pacific monsoon. Meteor. Atmos. Phys., 56, 57−79.

Irwin, R. P., and R. E., Davis, 1999: The relationship between the Southern Oscillation Index and tropical cyclone tracks in the eastern North Pacific. Geophys. Res. Lett., 20, 2251−2254.

Jury, M. R., 1993: A preliminary study of climatological associations and characteristics of tropical cyclones in the SW Indian Ocean. Meteorol. Atmos. Phys., 51, 101−115.

Jury, M. R., B. Pathack, and B. Parker, 1999: Climatic determinants and statistical prediction of tropical cyclone days in the Southwest Indian Ocean. J. Climate, 12, 1738−1746.

Kim, J.-H., C.-H. Ho, C.-H. Sui, and S.-K. Park, 2005: Dipole structure of interannual variations in summertime tropical cyclone activity over East Asia. J. Climate, 18, 5344–5356.

Kim, J.-H., C.-H. Ho, H.-S. Kim, C.-H. Sui, and S. K. Park, 2006: Systematic variation of summertime tropical cyclone activity in the western North Pacific in relation to the Madden-Julian oscillation. J. Climate, revised.

Kuleshov, Y., and G. de Hoedt, 2003: Tropical cyclone activity in the Southern Hemisphere. Bull. Austr. Meteorol. Soc., 16, 135–137.

Lander, M. A., 1994: An exploratory analysis of the relationship between tropical storm formation in the western North Pacific and ENSO. Mon. Wea. Rev., 122, 636–651.

Lander, M. A., 1996: Specific tropical cyclone track types and unusual tropical cyclone motions associated with a reverse-oriented monsoon trough in the western North Pacific. Wea. Forecasting, 11, 170–186.

Landsea, C. W., and W. M. Gray, 1992: The strong association between western Sahelian monsoon rainfall and intense Atlantic hurricanes. J. Climate, 5, 435–453.

Landsea, C. W., R. A. Pielke, Jr., A. M. Mestas-Nuñez, and J. A. Knaff, 1999: Atlantic basin hurricanes: Indices of climatic changes. Climatic Change, 42, 89–129.

Larson J., Y. Zhou and R. W. Higgins, 2005: Characteristics of landfalling tropical cyclones in the United States and Mexico: Climatology and interannual variability. J. Climate, 18, 1247–1262.

Lau, N. C., and M. J. Nath, 2003: Atmosphere–ocean variations in the Indo–Pacific sector during ENSO episodes. J. Climate, 16, 3–20.

Liebmann, B., H. H. Hendon, and J. D. Glick, 1994: The relationship between tropical cyclones of the western Pacific and Indian Oceans and the Madden-Julian oscillation. J. Meteor. Soc. Japan, 72, 401–411.

Maloney, E. D., and M. J. Dickinson, 2003: The intraseasonal oscillation and the energetics of summertime tropical western North Pacific synoptic-scale disturbances. J. Atmos. Sci., 60, 2153−2168.

Maloney, E. D., and D. L. Hartmann, 2000a: Modulation of hurricane activity in the Gulf of Mexico by the Madden-Julian oscillation, Science, 287, 2002−2004.

Maloney, E. D.., and D. L. Hartmann, 2000b: Modulation of eastern North Pacific hurricanes by the Madden-Julian oscillation, J. Climate, 13, 1451−1460.

Maloney, E. D., and D. L. Hartmann, 2001: The Madden-Julian oscillation, barotropic dynamics, and North Pacific tropical cyclone formation. Part I: Observations. J. Atmos. Sci., 58. 2545−2558.

463

Molinari J., and D. Vollaro, 2000: Planetary- and synoptic-scale influences on eastern Pacific tropical cyclogenesis. Mon. Wea. Rev., 128, 3296−3307.

Nakazawa, T., 1986: Intraseasonal variations of OLR in the Tropics during the FGGE year. J. Meteor. Soc. Japan, 64, 17−34.

Neumann, C. J., 1993: Global Overview. Chapter 1 of Global Guide to Tropical Cyclone Forecasting, edited by G. J. Holland, WMO/TC-No.560, Report No. TCP-31, World Meteorological Organization, Geneva.

Nicholls, N., C. W. Landsea, and J. Gill, 1998: Recent trends in Australian region tropical cyclone activity. Meteorol. Atmos. Phys., 65, 197−205.

Shapiro, L. J., 1987: Month-to-month variability of the Atlantic tropical circulation and its relationship to tropical storm formation. Mon. Wea. Rev., 115, 1598−1614.

Shapiro, L. J., 1989: The relationship of the quasi-biennial oscillation to Atlantic tropical storm activity. Mon. Wea. Rev., 117, 1545−1552.

Sobel, A. H., and E. D. Maloney, 2000: Effect of ENSO and the MJO on western North Pacific tropical cyclones. Geophys. Res. Lett., 27, 1739−1742.

Tang, B. H., and J. D. Neelin, 2004: ENSO Influence on Atlantic hurricanes via tropospheric warming. Geophys. Res. Lett., 31, L24204, doi:10.1029/2004GL021072.

Thorncroft, C., and K. Hodges, 2001: African easterly wave variability and its relationship to Atlantic tropical cyclone activity. J. Climate, 14, 1166–1179.

Wang, B., and J. C. L. Chan, 2002: How strong ENSO events affect tropical storm activity over the western North Pacific. J. Climate, 15, 1643–1658.

Wu, G., and N. C. Lau, 1992: A GCM simulation of the relationship between tropical storm formation and ENSO. Mon. Wea. Rev., 120, 958−977.

Xie, L., T. Yan, L. J. Pietrafesa, T. Karl, and X. Xu, 2005: Relationship between western North Pacific typhoon activity and Tibetan Plateau winter and spring snow cover. Geophys. Res. Lett., 32, L16703, doi:10.1029/2005GL023237.

464

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES Topic 4.2 : Possible Relationships Between Climate Change and Tropical Cyclone Activity Rapporteur: T. R. Knutson Geophysical Fluid Dynamics Laboratory/NOAA Princeton, NJ, 08542 USA Email: Tom.Knutson@noaa.gov Phone: 609-452-6509 Fax: 609-987-5063 Working Group: K. Emanuel, S. Emori, J. Evans, G. Holland, C. Landsea, K.-b. Liu, R. E. McDonald,

D. Nolan, M. Sugi, Y. Wang 4.2.1 Introduction This report reviews the current science on possible relationships between climate change and tropical cyclone (TC) activity/intensity on different time scales. Since variability of tropical cyclone activity/intensity on intraseasonal, interannual, interdecadal and multi-decadal scales is a topic addressed separately in another section of this report (Subtopic 4.1), the emphasis of this chapter will be on climate change (specifically meaning long-term trends in climate) as opposed to cyclical variations. An earlier review and assessment of this topic was presented in Henderson-Sellers et al. (1998). They concluded: i) there was no clear evidence for long-term trends in TC activity; ii) the potential intensity (PI) of storms would remain the same or increase by 10-20%, in terms of central pressure fall, for a doubling of CO2, although uncertainties remained with PI approaches; iii) little could be said about the future distribution of intensities or about future frequencies of TCs; and iv) the broad geographic regions of cyclogenesis and of occurrence of TCs were unlikely to change significantly. A 10-20% increase in central pressure fall would correspond to a smaller (roughly 5-10%) percentage increase in terms of maximum surface wind speeds. Walsh (2004) presents a more recent overview of the TC/climate change problem. The present report attempts to briefly summarize earlier work, but with a greater emphasis on work published since the Henderson-Sellers et al. assessment was completed, including recent published work on trends in observed TC metrics. A few statements in this report are non-peer-reviewed critiques of existing papers from our writing team. Since there was not agreement among the writing team about whether these types of critiques should be included in the report, we have identified such statements with square brackets []. 4.2.2. Background on tropical climate changes relevant to tropical cyclone activity There is substantial evidence that the large-scale environment in which hurricanes form and evolve is changing as a result of anthropogenic emissions of greenhouse gases and aerosols. A recent review of the climate change detection and attribution field (IADAG 2005) concluded that there is increasing evidence that “most of the global warming over the past 50 years is likely due to the increase in greenhouse gases.” A study of subsurface ocean data by Barnett et al. (2005) concluded that an anthropogenic warming signal is penetrating the world oceans, in broad agreement with model simulations that include the greenhouse gas forcing. Model-based attribution of early 20th century global warming (e.g. 1900-1944) to specific causes is more ambiguous, with various studies suggesting significant contributions from multiple factors, including increased greenhouse gases, solar variability, decreasing volcanic activity, and internal climate variability (e.g., Stott et al. 2000; Delworth and

465

Knutson 2000; Meehl et al. 2004; Knutson et al. 2006). On the regional scale of most relevance to local hurricane interaction, some aspects of the tropical climate appear to be changing in a trend-like fashion. There is increasing evidence that tropical sea surface temperature increases, as reported in recent studies (Emanuel 2005a; Webster et al. 2005), are at least partly a response to long-term increases in greenhouse gas concentrations. For example, Santer et al. (2006) find that observed SST increases in the Atlantic and North Pacific tropical cyclogenesis regions during the 20th century are unlikely to be due solely to unforced variability of the climate system, but are more realistically simulated in experiments using estimated historical climate forcing. Their internal climate variability assessment and external forcing results are made more robust by their use of 22 different climate models and two observed SST reconstructions. In the models in which individual forcing experiments were available, they find that the human-induced change in greenhouse gas forcing is the main cause of the 20th century warming, and particularly of the late 20th century warming. Their results support earlier regional surface temperature trend assessments based on a more limited set (two) of models (Knutson et al. 2006) or on a more limited set of forcings (Karoly and Wu 2005) both of which found model-based support for anthropogenically forced 20th century warming trends in the tropics and other regions. In Knutson et al., the simulations where anthropogenic and natural forcing agents were evaluated separately indicated significantly closer agreement with observed trends over much of the tropical oceans in the anthropogenic forcing runs than in the natural forcing or internal climate variability runs. The anthropogenic forcings in these experiments included changes in well-mixed greenhouse gases, ozone, and aerosols, as well as land use change, whereas natural forcings included solar variations and aerosols from volcanic eruptions. For the tropical North Atlantic, the roles of naturally occurring oscillations versus radiative forcing variability and trends on tropical Atlantic SSTs have also been evaluated using statistical modeling approaches. The potential importance of a naturally occurring large-scale oscillation of SSTs known as the Atlantic Multi-decadal Oscillation (AMO) was noted by Goldenberg et al. (2001), who found that multi-decadal variations in Atlantic major hurricane counts since the 1940s covaried with fluctuations in both Main Development Region (MDR) vertical wind shear as well as an AMO index derived from detrended SST data. Mann and Emanuel (2006) noted that late summer SSTs in the Atlantic Main Development Region (MDR) closely track, on long time scales, surface temperatures averaged over the entire Northern Hemisphere, with substantial warming over the 20th century. Using a statistical modeling approach, they suggested that most of the low-frequency (multi-decadal) variation and warming trend in MDR SSTs is being driven primarily by changing radiative forcing, as opposed to being part of the AMO. Their approach used terms proportional to global mean SST and to aerosol forcing, the latter of which they argued was justified by a regionally enhanced cooling response to aerosols over the tropical Atlantic during late summer. In general, climate forcing from aerosols is much more uncertain than the forcing due to increasing greenhouse gases (e.g., IPCC 2001). In another recent statistical analysis, Trenberth and Shea (2006) show that the method of construction of AMO indices can have a significant impact on AMO anomaly values for various time periods. They proposed that the index be constructed as a residual after removal of a near-global (60N-60S) SST component, as opposed to residual from a linear trend (as in Goldenberg et al. 2001). Using this approach, they derive a revised AMO index with smoothed anomaly values of about +/- 0.2 C and a transition from negative to positive values in the mid 1990s. However, the contribution of their low-pass-filtered AMO anomalies to the record summer of 2005 values is quite small (<0.1C), and the anomalies from 1870 to 1900, are also much smaller (closer to zero) compared to those using the method of Goldenberg et al. [In removing the global or near-global mean SST from the Atlantic SST series, both Mann and Emanuel (2006) and Trenberth and Shea (2006) include the Atlantic SST in their computation of the global or near-global mean. This procedure could have the effect of artificially damping the AMO amplitude.] Enfield and Mestas Nuñez (2000) have previously published a means of deriving an “Atlantic Multidecadal Mode” based not on linear trend removal, but on a complex empirical orthogonal function

466

(CEOF) decomposition, in which a “Global Warming Mode” is distinguished from the Atlantic Multidecadal and Pacific interdecadal modes based on the CEOF modal decomposition. In their analysis the AMO is the third CEOF in a dataset from which an ENSO-related CEOF had previously been removed (i.e., in practical terms, the AMO is their fourth CEOF). It should be noted that different SST reconstructions have been used by various investigators, which may also contribute to differences seen in the resulting analyses (e.g., Santer et al. 2006). The existence of a robust AMO-like internal mode of the climate system is supported by some climate models, which simulate internal modes of variability that resemble the observed AMO signal in several respects (Delworth and Mann 2001; Knight et al. 2005) Model based studies also indicate that such multidecadal variations of Atlantic SSTs can have important impacts on vertical wind shear in the Atlantic MDR (Zhang and Delworth, 2006; Knight et al. 2006), which Goldenberg et al. (2001) proposed can then affect Atlantic hurricane activity. Elsner (2006) uses Granger causality statistical analysis to demonstrate that global mean temperature can be used to predict North Atlantic SST but not the other way around. This, he argues, supports the hypothesis that greenhouse gases are the causal forcing agent for global temperatures and thus for North Atlantic SSTs and hurricanes. To eliminate nonstationarity in the data, Elsner time differences both the global temperature and North Atlantic SST series before performing the causality tests. [Time differencing acts as a strong high-pass filter on the data, and thus his conclusions about Granger causality strictly apply only to the higher frequency fluctuations that remain in the data. The inference that global temperature Granger causes North Atlantic SST fluctuations on multidecadal time scales thus depends on the assumption that the direction of causality for the high frequency fluctuations also applies to the lower (filtered) frequencies, which has not yet been demonstrated by his analysis.] Trenberth et al. (2005) have reported a substantial increase (1.3% +/- 0.3% per decade) in column-integrated atmospheric water vapor over the global oceans (1988 to 2003) as derived from the special sensor microwave imager (SSM/I) satellite data set. Thus, it appears that tropical precipitable water vapor is increasing in a manner consistent with the notion of approximately constant relative humidity, and in accord with model simulations of tropical relative humidity under warming conditions (e.g., Knutson and Tuleya 2004). As noted by Trenberth et al., the relatively short available record in their study is a limitation with regard to inferring an anthropogenic climate change signal. The vertical profile of historical tropospheric temperature trends in the tropics has been a subject of considerable debate in the climate change community. In a recent study, Santer et al. (2005) examined the profile of temperature changes for the period 1979-1999 produced by a large ensemble of climate models, all incorporating a range of historical forcings including greenhouse gases and aerosols, with some of the models incorporating volcanic eruptions. The climate models generally simulate an enhanced warming of the tropical upper troposphere relative to the surface. In contrast, the observed vertical profile of radiosonde-derived temperature trends over this period has a distinctly different character from the model simulations, with the observations showing much smaller tropospheric warming trends relative to the surface. Finally, Santer et al. showed that both models and observations have interannual variations in upper tropospheric temperatures that are enhanced relative to the surface variations. Thus the vertical structure of interannual variations is similar to that of modeled trends (1979-1999), but is in sharp contrast with the vertical structure of observed trends (1979-1999) in tropical tropospheric temperatures. Their results are suggestive of serious remaining problems with radiosonde-derived and satellite-derived temperature trends--a conclusion also receiving some support from two other recent studies which examined issues with radiosonde-based observations (Sherwood et al. 2005) and satellite-based analyses (Mears and Wentz 2005). The possibility that tropospheric trend estimates from radiosonde-based observations (including reanalyses) may be unreliable should be considered as a caveat when reviewing other published reports on related trend measures. Other measures of tropical climate relevant to hurricane formation which have been examined for possible trends include CAPE and potential intensity. For example,

467

Gettleman et al. (2002) found a preponderance of upward trends in tropical CAPE since roughly the early 1960s. DeMott and Randall (2004) examined a larger number of tropical stations over a shorter period (1973-1999) and reported a more evenly divided mixture of increasing and decreasing CAPE trends. Trenberth (2005) questioned the reliability of the radiosonde data in DeMott and Randall’s larger sample. Free et al. (2004), using a selected set of 14 tropical island radiosonde stations, found only small, statistically insignificant trends in potential intensity over the periods 1975 to 1995 and 1980 to 1995. Emanuel (2006) reported a 10% increase in Atlantic MDR potential intensity since 1982 based on HadISST and NCEP reanalysis data. Concerning large-scale circulation indices, Bell and Chelliah (2006) have statistically linked multi-decadal changes in Atlantic hurricane activity to a series of large-scale circulation features, all correlated to a multi-decadal circulation signal derived from empirical orthogonal function (EOF) analysis of upper tropospheric velocity potential. The indices of Bell and Chelliah’s modes are of insufficient length to determine whether they have a cyclical or trend-like character, or some combination of trend and cycle. No published studies to date have linked pronounced historical circulation changes such as Bell and Chelliah’s multi-decadal signal or Goldenberg et al.’s (2001) index of Atlantic MDR vertical shear to a radiative forcing mechanism. Similarly, Saunders and Lea (2005) and Elsner et al.(2000; 2006) find statistical links between U.S. landfalling hurricane activity and large-scale circulation anomalies, although long-term climate trends in their predictors have not been firmly established. In the Southern Hemisphere, observed trends in the Southern Annular Mode have been at least partly attributed to anthropogenic forcing (Marshall et al. 2004). Pezza and Simmonds (2005), commenting on the rare atmospheric conditions associated with the first reported hurricane in the South Atlantic (Catarina, 2004), suggested that observed and predicted trends in the Southern Annular Mode could increase the probability of such conditions in the future. Two recent studies (Rotstayn and Lohmann 2002; Held et al. 2005) have found that 20th century trends in Sahel rainfall may have been at least partially forced by anthropogenic forcing, albeit through different physical mechanisms in the two studies. In Rotstayn and Lohmann’s study, anthropogenic indirect aerosol forcing through the interaction of sulfate aerosol with cloud and precipitation processes, causes a pronounced decrease in rainfall in the Sahel, whereas in Held et al. (2005) the Sahel drying from the 1950s to the 1980s is simulated by approximately equal contributions of internal climate variability and radiative forcing (the latter being primarily anthropogenic “direct effect-only”aerosol forcing and increasing greenhouse gases). Vecchi et al. (2006) report evidence for a century-scale weakening trend in the Walker Circulation in the Pacific, similar to that which occurs during El Niño, and consistent with the slightly El Nino-like warming trends predicted by some climate models. Both West African monsoon activity and El Niño have been statistically linked to Atlantic hurricane activity (e.g., Gray 1990; Bell and Chelliah 2006). The Southern Annular Mode, Pacific Walker circulation, and Sahel drought studies cited above serve as reminders that the relationship between radiative climate forcing and hurricane response may involve a variety of complex tropics-wide or even global-scale phenomena. 4.2.3 Observed trends and low-frequency variability of tropical cyclone activity In the past year, two observational studies of low-frequency variability and trends in several measures of tropical cyclone activity (Emanuel 2005a; Webster et al. 2005) have attracted considerable attention in the hurricane research community. Emanuel (2005a) developed a “Power Dissipation Index” (PDI) of tropical cyclones, based on the time integrated cube of the estimated maximum sustained surface wind speeds, some of which are inferred from central pressure reports, for the Atlantic and Northwest Pacific tropical cyclone basins from the late 1940s to 2003. After adjusting for time-dependent biases due to changes in measurement and reporting practices, Emanuel reported an approximate doubling of the PDI over the period of record, with contributions from apparent increases in both intensity and mean storm duration. The low-pass

468

filtered PDI series in his study were significantly correlated with large-scale tropical SST indices for both basins. Pielke (2005) noted that there are no evident trends in observed damage in the North Atlantic region, though Emanuel (2005b) notes that a PDI series such as Landsea's (2005) based on only U.S. landfalling data, contains only about 1 percent of the data that Emanuel's (2005a) PDI contains, which is based on all storms over their entire lifetimes. Thus a trend in basin-wide PDI may not be detectable in U.S. landfalling PDI since the former index has a factor of 10 advantage in signal to noise ratio. A subsequent comment by Landsea (2005) resulted in adjustments, particularly removing much of the large post-2000 upswing in the Atlantic PDI series through 2003, although Emanuel (2005b) reported that these adjustments had minimal impact on the Northwest Pacific results. In particular, Landsea’s (2005) PDI analysis for the Atlantic basin shows no evidence for a trend from 1949-2004, similar to time series of major hurricane counts or Accumulated Cyclone Energy (ACE, Bell and Chelliah 2006), provided that wind speeds early in the record are not adjusted, as Landsea (2005) now recommends. Landsea’s (2005) PDI for U.S. landfalling Atlantic tropical cyclones (1900 to 2004) also shows no evidence of an upward trend, with the past two seasons (2004 and 2005) being strong positive outliers, with similar magnitude to that estimated for 1886. As noted by Emanuel (2005b), this is not necessarily inconsistent with the observed trends in basin-wide statistics. Recently Emanuel (2006) presented an alternative PDI measure for the Atlantic basin, the storm maximum PDI, which was extended back into the late 1800s. This measure tracks the long-term variation in Atlantic MDR SST—particularly the century-scale warming trend--fairly closely, and is also notably well-correlated with MDR SST, particularly after 1970 (r2 = 0.83 from 1970 onwards, using low-pass (1-3-4-3-1) filtered data). A similar behavior is seen for TC counts for the Atlantic basin (e.g., Mann and Emanuel 2006). The reliability of Atlantic basin-wide TC measures prior to the 1940s is highly debatable, as Landsea (2005) argues that the PDI values even from the more recent 1940s to the 1960s are likely to be substantially undercounted due to lack of routine aircraft reconnaissance and geostationary satellite monitoring of TCs far from land. Mann and Emanuel (2006) argue that detection of the existence of TCs in the years prior to the 1940s was less problematic than TC intensity estimates, since in the absence of aircraft and satellites based guidance to warn them off, ships often encountered TCs at sea, at least peripherally. On the other hand, Landsea et al. (2004) had earlier estimated the number of “missed” Atlantic basin tropical storms and hurricanes per year to be on the order of 0-6 for the period 1851-85 and 0-4 for the period 1886-1910. They argued that the TC record over the Atlantic should by no means be considered complete for either frequency or intensity of tropical storms and hurricanes for the years 1851 to 1910, in contrast to the more complete and accurate information available for landfalling TCs along much of the U.S. coastline. In a more regionally focused study, Mock (2004) analyzed records of TC activity from 1769 to 2003 for the state of South Carolina in an effort to assess a relatively homogeneous multi-century record of tropical storm and hurricane strikes. This analysis suggests pronounced multidecadal variability, but no long-term trends. Given the well-established communities along the South Carolina coastal regions since the 18th century, it is unlikely that any significant hurricanes were not captured in this record. Webster et al. (2005) reported that the number of category 4 and 5 hurricanes has almost doubled globally over the past three decades. Although their analysis spans a shorter time period than Emanuel's, due to their decision to limit the analysis to the satellite era, their results indicate that a substantial increase has occurred in all six tropical storm basins. They found no trend in the numbers of tropical storms and hurricanes or in the maximum wind speed observed globally each year. While they did also find an increasing trend in the duration of Atlantic tropical cyclones over this period, no significant trend was identified in the remaining global basins for duration. In a follow-on study, Hoyos et al. (2006) found that the increasing trends in category 4 and 5 tropical cyclones are principally correlated with SST as opposed to other environmental factors.

469

The recent Emanuel and Webster et al. studies continue to be the subject of much debate in the hurricane research community, particularly with regard to homogeneity of the tropical cyclone data over time and the required adjustments. For example, Knaff and Sampson (2006) reanalyzed the maximum intensities of Northwest Pacific tropical cyclones over the period 1966-1987 and found a much reduced upward trend in annual numbers of category 4 and 5 storms in their reanalyzed data relative to the original best track data. Chan (2006) extended the analysis of Webster et al. for the Northwest Pacific basin back to earlier years and argued that the “trend” in that basin is part of a large interdecadal variation (see also Webster et al. 2006; Chan and Liu 2004). Chan used unadjusted data from the earlier part of the record, in contrast to the adjustments for this period proposed by Emanuel (2005a) for the basin. Landsea et al. (2006) propose that much – perhaps the majority – of the global increase in Category 4 and 5 TCs since 1970 may be due to data reliability issues in that strong TCs are more accurately monitored in the more recent years. They documented six additional Category 4 and 5 TCs in the North Indian Ocean during the 1970s and 1980s, which were not counted as such in the Webster et al. (2005) study. The inclusion of these extreme TCs make the trend found in the North Indian Ocean much weaker, perhaps insignificantly so. They argue that such systematic undercounts are endemic in the global TC records, especially in basins that rely primarily upon satellite imagery for intensity monitoring (that is, all but the Atlantic). Using a different approach, Sriver and Huber (2006) computed power dissipation statistics from ECMWF (ERA-40) reanalysis data from 1958 to 2001. Despite the coarse resolution of the reanalysis data (1.125o longitude by 1.125o latitude), their resulting global indices, normalized by their standard deviations, are well-correlated with Emanuel’s (2005a) Atlantic + Western North Pacific PDI, particularly after 1978. Sriver and Huber estimated a sensitivity of global power dissipation of roughly 60% per 0.25 degree Celsius SST increase. [The ERA-40 reanalysis has benefited from improvements in the observing systems over the years, which conceivably could have led to inhomogeneities or artificial increasing trends in the PDI measures derived by Sriver and Huber, particularly considering trends involving the pre-1979 era, when the agreement with Emanuel’s PDI is less compelling. For example, Special Sensor Microwave/Imager (SSM/I – 1987), European Remote Sensing Satellite (ERS – 1991) and increased cloud motion winds (1970s through 1990s) should contribute to better defined surface winds in the tropics and thus may cause some artificial increasing trend in ERA-based PDI values.] Of possible relevance to the tropical storm issue, Hoskins and Hodges (2005) discuss problems with using the ERA40 and other reanalyses for examining past Southern Hemisphere extratropical cyclone numbers and intensity. Michaels et al. (2006) hypothesized that Atlantic tropical cyclones respond to an SST threshold such that major hurricanes are possible only for storms experiencing SSTs above the threshold value at some point in their lifetime. Using a statistical-analog approach, they infer an intensity sensitivity of about 6.3% in wind speed for a 2 degree Celsius SST rise. In comparing these sensitivities, note that PDI depends on the cube of the wind speed and includes effects of storm duration and frequency. Nonetheless, a much higher intensity sensitivity is likely implied by Sriver and Huber’s analysis than by Michaels et al. For example, if we assume based on Emanuel (2005) that the PDI change is half due to intensity increase and half to duration increase with no frequency change, the above sensitivity estimates still differ from each other by more than a factor of 10. Over the period 1986-2005, Klotzbach (2006) finds no significant change in global net tropical cyclone activity, and a small trend (~+10%) in category 4 and 5 TC frequencies. He restricted his analysis to this 20-yr period owing to data quality concerns. In particular, while he finds a large increase in TC activity in the Atlantic from 1986-2005, there is a nearly commensurate decrease in the Northeast Pacific, and the remaining global basins show negligible changes in the 20 year period. Klotzbach's Fig. 2 shows a tropical SST warming trend of roughly 0.2oC during this period. It should be noted that climate change detection studies in a variety of contexts have found that the ability to detect significant trends in climate records is reduced as the record length is shortened, and Klotzbach’s analysis is

470

based on a relatively short record compared with other analyses. On the other hand, Klotzbach’s conclusion that there have been only minor alterations in global TC activity in the last 20 years is consistent with Landsea et al.’s (2006) assertions that monitoring changes in the 1970s and 1980s, rather than true climate changes, may be responsible for the increased extreme TC occurrence during from 1990s and early 2000s found in Webster et al. (2005). Klotzbach’s analysis included a portion (1986 to 1990) of the period whose intensity estimates were questioned by Landsea et al. (2006). Using a partial correlation statistical analysis, Elsner et al. (2006) examined the relationships between global temperature, tropical Atlantic SST, and Atlantic PDI on high-frequency interannual time scales. They concluded that the positive influence of global temperature on PDI was limited to an indirect connection through the tropical Atlantic SSTs. After controlling for the effect of tropical Atlantic SSTs on PDI, the correlation of PDI with global temperatures was slightly negative. This result was consistent with idealized modeling studies (Shen et al. 2000) and with statistical analyses of ENSO-Atlantic TC relationships (Tang and Neelin 2004), both indicating inhibiting effects of tropospheric stabilization on TC intensity or frequency. Kamahori et al. (2006) examine how the records of typhoon days compare between the Japanese Meteorological Agency (JMA) typhoon best tracks and those from the Joint Typhoon Warning Center (JWTC, which were used in Emanuel (2005a) and Webster et al. 2005)) from 1977 until 2004. They found a 15-30% increase in TC days with an intensity of category 2 or higher in both data sets, although with pronounced differences between the two data sources as to the distribution of storms within the category 2-5 range. For example, they found that the number of Category 4 and 5 typhoon days decreased from 7.2 per year in 1977-90 to 4.3 per year in 1991-2004 in the JMA database. This contrasts with the JWTC dataset, which showed for Category 4 and 5 typhoon days 9.8 per year from 1977-90 and 16.9 per year from 1991-2004. Undoubtedly, the discrepancy relates to JMA vs JWTC satellite treatment of TC intensities once aircraft reconnaissance was discontinued there in 1987. There is currently no guidance as to which dataset is more reasonable in assessing true extreme TC climate trends. Some regional trends in prevailing typhoon tracks in the western North Pacific for the period 1965 to 2003 have been reported by Wu et al. (2005), although they were not able to distinguish between anthropogenic impacts or long-term natural variability. The changes in tracks were found to be consistent with expected changes based on large scale circulation (steering flow) changes. 4.2.4. Paleoclimate proxy studies of past TC activity Paleotempestology is the term for an emerging field of science that attempts to reconstruct past tropical cyclone activity using geological proxy evidence or historical documents. This work attempts to expand knowledge about hurricane occurrence back in time beyond the limits of conventional instrumental records, which cover roughly the last 150 years. A long-term record of hurricane activity on timescales of centuries to millennia is especially important in understanding the spatial and temporal variability of the rare but most intense landfalling hurricanes like Camille (1969) or Andrew (1992), which may have return periods of longer than 150 years. A broader goal of paleotempestology is to help researchers explore physically based linkages between prehistoric TC activity and other aspects of past climate. This would provide important independent evidence for causes of changes in hurricane activity, and possibly assist in understanding the potential for future climate changes to affect hurricane activity, or vice versa (e.g., Section 7). Among the geologically based proxies, overwash sand layers deposited in coastal lakes and marshes have proven to be quite useful (Liu and Fearn, 1993, 2000; Liu 2004; Donnelly and Webb 2004). The storm surge plus wave run-up during an intense hurricane can overwash a beach barrier, eroding sand and depositing a layer of the eroded sand material in a lake or marsh behind the barrier. These layers

471

can then form a stratified record through time of intense storm overwash events. Cores of these layers can be retrieved and the layers analyzed in terms of their thickness, composition, age, and frequency of occurrence. The age of deposits is estimated through various radiometric dating techniques applied to the surrounding organic matter, supplemented by other information. By comparison of the characteristics of these deposits with those of well-observed storms in the historical record, inferences about past storm events can be made. For example, Liu and Fearn (1993; 2000) have constructed a 5,000-year paleo record of inferred category 4 and 5 hurricane strikes in Alabama and northwestern Florida using this technique. Similar methods have been used to produce proxy records of hurricane strikes from back-barrier marshes in Rhode Island and New Jersey extending back about 700 years (Donnelly et al. 2001a, 2001b; Donnelly et al. 2004; Donnelly and Webb 2004), and more recently in the Caribbean (Donnelly 2005). In interpreting these records, long-term changes in sea level must also be taken into account. The frequency of cyclone or “super-cyclone” occurrence in the Australia region over the past 5000 years has been inferred from chronostratigraphic series of shelly beach ridges (Nott and Hayne, 2001; Hayne and Chappell 2001). Stable isotope signals in tree rings (Miller et al. 2003), cave deposits (Frappier et al. 2006) and coral reef materials are also being actively explored for their utility in providing paleoclimate information on tropical cyclone activity. These methods attempt to exploit an oxygen isotope signal that distinguishes rain originating in hurricanes from that in low-latitude thunderstorms (Lawrence and Gedzelman, 1996). Rainwater from a hurricane is eventually incorporated into the tree-ring, cave deposit, or reef material where it may be preserved as a long-term proxy record. The above studies are beginning to show some promise as a method of inferring aspects of past tropical cyclone activity. Historical documents apart from traditional weather service records can also be used to reconstruct some aspects of past tropical cyclone activity. For example, investigators have used sources such as newspapers, plantation diaries, various instrumental weather records, and annals in the Carribbean to reconstruct past tropical cyclone activity in the U.S., Caribbean, Gulf of Mexico, and Atlantic basin for up to several centuries before present (Ludlam, 1963; Millas, 1968; Fernandez-Partagas and Diaz, 1996; Chenoweth, 2003; Mock 2004). Spanish and British historical archives may be a useful source for further investigation (Garcia Herrera et al. 2004; 2005). Even longer documentary records of tropical cyclone activity, extending back for more than 1000 years, have been found and investigated in China (Liu et al. 2001; Louie and Liu 2003; Louie and Liu 2004). Paleoclimate researchers are continuing to investigate these multiple sources of information on pre-historic tropical cyclone activity, and to validate where possible, the paleoclimate proxy records against hurricane observations from the more recent, well-observed part of the historical record. These studies should provide an increasingly useful independent source of information on the tropical cyclone-climate connection, as well as a better-constrained long-term perspective on hurricane risk from rare but extreme hurricanes. Future efforts will include expansion of geographical coverage, development of new proxies, coupling of multiple proxy sources, improved calibration, and integration with modeling and advanced statistical techniques. 4.2.5. Use of theory and models to understand past variations in tropical cyclone activity Theory and models of tropical cyclone activity may provide useful information both on the interpretation of past changes in activity and on possible future changes due to such factors as greenhouse gas-induced global warming or natural climate variability. In this section, the utility of the theoretical and modeling approaches is assessed based on analyses of climatological (seasonal) variations and past interannual to interdecadal variations and trends. The theories include potential intensity theories as well as empirical indices which attempt to relate tropical cyclone frequency to large-scale environmental conditions. The models range from global climate models to higher resolution regional models aimed at simulating hurricane structure in more detail.

472

a) Assessment of simulated TC climatologies and seasonal cycles Extended integrations of global climate models in principle allow for an assessment of the frequency, intensity, duration, structure, and tracks of tropical cyclone-like features in the model. In practice, simulation of realistic intensities and detailed structures of the TC’s is hampered by the coarse resolution generally required of such global models, as discussed below. In addition, the fidelity of the global model’s TC genesis process compared to the real world has not been well established. In the global models, TCs are located and tracked in model data using objective techniques that are usually based on a local maximum of cyclonic relative vorticity at 850 hPa and often involve other criteria such as: life-time; warm core; maximum winds above a threshold; and local minimum MSLP (e.g. Tsutsui and Kasahara 1996; Vitart et al. 1997; Sugi et al. 2002; McDonald et al. 2005). The location and tracking method tends to be unique to each study which makes it difficult to compare the results of the different studies directly. Walsh et al. (2006) provide recommendations for providing homogeneous comparisons of various resolution models for determining tropical cyclone frequencies. They base this upon an analysis of how minimal tropical storms (with maximum winds at 17.5 m/s) would be depicted under various resolutions. Use of such resolution-based criteria for determining tropical cyclone occurrence should allow for more rigorous quantitative comparisons of global (and regional) climate model output of tropical cyclones frequencies. The global climate models used for tropical cyclone analysis have tended to be of low (300km) horizontal resolution (e.g. Vitart et al. 1997; Tsutsui 2002; Bengtsson et al. 2006) or of medium (120km) resolution (e.g. Sugi et al. 2002; McDonald et al. 2005; Hasegawa and Emori 2005; Yoshimura and Sugi 2005). The grid-scale of the low and medium resolution models is larger than the typical scale of tropical cyclones and this can lead to a poor simulation of tropical cyclones (e.g. Vitart et al. 1997; McDonald et al. 2005). The cyclones tend to have a larger horizontal scale, and although they have warm cores, the intense inner core is not well-simulated. Thus, the cyclones have lower wind speeds than observed tropical cyclones (Vitart et al. 1997). Minimum central pressures tend to be better simulated than the maximum surface wind speeds. Recent studies studies have used higher resolutions of 50km (Chauvin et al. 2006) and 20km (Oouchi et al. 2006), but models of this resolution are too expensive for most modeling centers to use for long climate change experiments. An alternative approached is to use a global model with a stretched grid (i.e. higher resolution) over the region of interest (e.g. Chauvin et at. 2006) although this limits the study to the region where the resolution is high. Even at that resolution, the highest simulated TC intensity reported by Oouchi et al. was about 932 hPa, compared with the observed record of 870 hPa, indicating the limitation of their global model in simulating very intense TCs. More realistic maximum intensity levels, including their geographic distribution in the NW Pacific basin, have been simulated by downscaling individual storm cases from a coarse-grid global model into an operational regional high-resolution hurricane prediction system (Knutson et al. 1998) or into a regional climate model (Walsh and Ryan 2000; Walsh et al. 2004). Even though smaller-scale features of the individual cyclones are typically not well simulated in the global models, these models are able to reproduce some aspects of the observed climatology and inter-annual variability of tropical cyclones (Tsutsui and Kasahara 1996; Sugi et al. 2002; Camargo et al. 2005; McDonald et al. 2005). Most models are able to simulate tropical cyclone-like disturbances in roughly the correct location and at the correct time of year, although all models exhibit some biases. Several models simulate tropical cyclones in the South Atlantic (Vitart et al. 1997; Sugi et al. 2002; McDonald et al. 2005; Oouchi et al. 2006) where they are rarely observed (Pezza and Simmonds 2005), although not all models simulate storms there (Camargo et al. 2005). The global models’ simulated TC tracks are sometimes shorter than observed (Tsutsui and Kasahara

473

1996; Sugi et al. 2002; Camargo et al. 2005) but those in the high resolution model of Oouchi et al (2006) are better simulated, whereas those in the low resolution models ECHAM3 and ECHAM4 are too long (Camargo et al 2005). Some of these differences in length may be due to the objective techniques used to identify and track the cyclones in the model data. The simulated global annual frequency of TCs in global models varies, with some models simulating too many TCs (e.g., McDonald et al. 2005) while others simulate too few (Camargo et al. 2005). Both from comparing results between different models (e.g., Carmago et al. 2005) and from sensitivity experiments with a given model (Vitart et al. 2001; Emori et al. 2005), it is evident that both model resolution and model physics can play important roles in determining the frequency of TC occurrence in the global models. While models in many existing studies have demonstrated an ability to simulate many aspects of the seasonal variability of tropical cyclone frequency in each basin, all of the models have some errors in both frequency and timings. These errors are basin-, season- and model-dependent. Increasing the horizontal resolution of the global models typically improves the simulation of the individual cyclones (Bengtsson et al. 1995) but may not improve the tropical cyclone climatology and interannual variability, as it is also important that the models have a good simulation of the large-scale circulation for the latter. Tropical cyclone occurrence is observed to be correlated to the phase of ENSO (Chan 1985). This implies that climate models also must simulate realistic ENSO and decadal variability under present day and future climate conditions as a necessary condition for providing reliable future projections of TC activity in these regions (e.g. Nguyen and Walsh 2001; Chan and Liu 2004). Chan and Evans (2002) examine CCM3 and GISS ensemble simulations of the structure of the East Asian summer monsoon in the present climate. Since tropical cyclogenesis in the western North Pacific is dominated by the monsoon, realistic simulations of the monsoon and is variability are necessary for accurate representation of genesis. Chan and Evans demonstrate that there is more variability of monsoon behavior among individual members of the ensembles than between ENSO extremes. An alternative approach to explicit global model simulation is to use an empirical “seasonal genesis parameter” (e.g. Ryan et al. 1992; Watterson et al. 1995) to infer a genesis frequency from climate model data (Tsutsui and Kasahara 1996; Royer et al. 1998; McDonald et al. 2005; Chauvin et al. 2006). Great caution is required when applying a parameter developed for present day climate to future predictions as the statistical relationships may not be valid under altered climate conditions (Ryan et al. 1992). Royer et al. (1998) and Emanuel and Nolan (2004) (see also Nolan et al. 2006) have proposed a refined versions of Gray’s (1979) genesis index that avoid the use of factors such as threshold SSTs that themselves may well vary in an altered climate (e.g., Henderson-Sellers et al. 1998). These methods typically produce plausible maps and seasonal cycles of TC genesis. Recent efforts have begun in assessing performance of the Emanuel and Nolan scheme with regard to interannual (ENSO) variability (Carmargo et al. 2006). Wu and Wang (2004) showed that TC track climatologies can be inferred from global model data using the climatological mean velocity fields as input to a TC trajectory model. In an initial application of this technique in the Northwest Pacific, Wu and Wang demonstrated that the main characteristics of the current climatology of TC tracks can be reproduced using this approach with the mean circulation data from NCEP-NCAR reanalyses. Since any systematic changes in TC tracks could have important impacts on TC-related damage, this approach provides an alternative means of exploring potentially important climate-change-induced TC impacts. The present-climate performance of theoretical frameworks such as hurricane potential intensity theory has been assessed to some degree based on geographical or seasonal variations of real-world tropical cyclone intensities (e.g., Emanuel, 1987; Emanuel 2000; Holland 1997; Tonkin et al. 2000). These assessments show that the theories have some skill at hindcasting seasonal and geographical variations of maximum intensities similar to those in the real world, although shortcomings can also be

474

identified (e.g., Tonkin et al. 2000). Camp and Montgomery (2001) and Holland (1997) have noted that the Emanuel and Holland potential intensity theories have somewhat different sensitivities to certain environmental factors, such as relative humidity. The current climate-based assessments to date suggest that potential intensity theories are a plausible means of relating real-world intensities of TCs to the large-scale environmental thermodynamic conditions in which the storms form and develop. The capability of models to simulate realistic intensities can also be assessed based on experience with operational TC prediction. Using a simplified numerical modeling framework, Emanuel (1999) showed several cases of successful hindcasts of hurricane intensity, derived from each storm’s initial intensity, the large-scale atmospheric thermodynamic environment (which also determines the potential intensity), and a representation of ocean mixing (potentially creating a cool wake) beneath the storm. Knutson and Tuleya (2004) used an idealized hurricane model derived from an operational hurricane prediction system to assess possible impacts of climate change on hurricane intensity in the absence of wind shear effects. The operational performance of the GFDL hurricane model for intensity prediction and the relevance of its performance for the climate change/hurricane intensity problem has been a subject of debate (Michaels et al. 2005; Knutson and Tuleya 2005). As noted by Knutson and Tuleya (2004) their results should be interpreted as analogous to potential intensity. Emanuel et al. (2006) recently introduced a new statistical-deterministic TC simulation approach based on combination of synthetically generated storm tracks and a TC intensity prediction framework which is applied along the path of the TC. The TC intensity scheme includes the impact of the large-scale thermodynamic environment, along with representations of wind shear and ocean interaction effects, through a simple axisymmetric balance hurricane model coupled to a one-dimensional ocean model. Being highly simplified and computationally efficient compared with full three-dimensional atmosphere/ocean models, the approach has been used to generate TC statistics based on thousands of synthetic tracks, under various assumptions about genesis locations. The method produces a reasonable histogram of maximum wind speed occurrence for the Atlantic basin as a whole, and to some degree for two U.S. coastal cities that were examined (Miami and Boston). This method was then applied by Emanuel (2006b) to controlled climate changes. Using a sample of 3000 synthetic Atlantic storm tracks and holding the tracks themselves fixed, Emanuel re-ran the intensity model with 10% spatially uniform increases, in succession, in potential intensity, vertical wind shear, and ocean mixed layer thickness. These led, respectively, to changes in net tropical cyclone power dissipation of +65%, -12% and +4%. b. Assessment of simulated interannual to decadal variability of TC activity The interannual variability of TC occurrence in global models can be tested by comparing cyclones simulated in models forced with observed SSTs to tropical cyclone observations from the same period (Tsutsui and Kasahara 1996; Vitart et al. 1997; Sugi et al. 2002; McDonald et al. 2005; Camargo et al. 2005). The nine-member ensemble of Vitart et al. 1997 and the 40 yr experiments used by Camargo et al. (2005) are better suited for analysis of the inter-annual variability than are the shorter experiments used by Tsutsui and Kasahara (1996), Sugi et al. (2002) and McDonald et al. (2005) because of the larger sample sizes. The correlation of the global annual number of tropical cyclones with the observed varies from 0.15 in the JMA model (Sugi et al. 2002) to 0.41 in GFDL model (Vitart et al. 1997). The correlations are better in some seasons, basins and models than in others. The correlations tend to be highest in the west North Pacific and North Atlantic basins (Vitart et al. 1997; Carmago et al. 2005), possibly because of the importance of ENSO in those regions. The interannual variability performance of the 20 km grid global model of Oouchi et al. (2006) has not yet been assessed. In a coarse-grid global model investigation of interdecadal variability of tropical storm occurrence in the Atlantic, Vitart and Anderson (2001) were able to simulate a decrease in tropical storm frequency for the 1970s in comparison to the 1950s, similar to observations, by specifying the observed SST changes for the globe (and specifically for the tropical North Atlantic) in their model. The decreased frequency in their model was linked to increased vertical wind shear and reduced CAPE in the tropical

475

storm formation regions. A correlation of hurricane activity with tropical vertical wind shear has also been noted in observational studies of Atlantic TC variability (e.g., Goldenberg et al. 2001; Bell and Chelliah 2006). c. Assessment of observed trends in TC activity and related measures The substantial increases in tropical cyclone Power Dissipation Indices (PDIs) reported by Emanuel (2005a; 2006) and the reported increases in the numbers and percentages of TCs attaining category 4 or 5 intensity (Webster et al. 2005; Hoyos et al. 2006) raise the question of whether these changes can be reconciled with existing theory or modeling work. A caveat to such comparisons is that existing modeling work, such as Knutson and Tuleya (2004) is highly idealized in terms of both climate forcings (CO2 only vs a mixture of known historical forcings) and in neglect of potentially important factors such as vertical wind shear. Therefore, such existing studies can only provide a rough guide as to expected responses of hurricanes to both past and future climate changes. Emanuel (2005a; 2006) provided some discussion of the discrepancy between observed and theoretical results, in particular noting the importance of considering potential intensity and not simply SST in attempting such comparisons. Emanuel (2005a) noted that the doubling of the North Atlantic plus western North Pacific PDI in the last 30 years implied in his analysis was due to increases in both the accumulated annual duration of storms and the peak intensities of TCs. The annual average storm peak wind speed summed over the Atlantic and North Pacific basins (in terms of V3) increased by about 50% during the period, implying roughly 30% per degree Celsius sensitivity of wind intensity. Emanuel (2005a) reported that the potential intensity as estimated from atmospheric re-analysis data had increased by about 10% rather than the predicted 2-3% over the period, owing to the failure of atmospheric temperature warming to keep pace with the SST warming. Emanuel noted that predicted peak intensity sensitivity to SST from theory was only about 5% per degree Celsius, which, for an SST increase of 0.5oC, implied a PDI increase of only about 8-12%. In an analysis of more recent (post-1979) Atlantic basin data Emanuel (2006) suggested that the 10% increase in potential intensity in that region was partly attributable to decreases in mean surface wind speeds over the basin as well as SST increases of about 0.5 oC, implying about 20% per degree Celsius sensitivity to SST alone. The modeling study of Knutson and Tuleya (2004; 2005) found a peak wind speed sensitivity of about 3.3% per degree Celsius (or 3.7% per degree Celsius if maximum winds are inferred from central pressures following Landsea 1993). Thus while highly idealized, their results imply a discrepancy of roughly a factor of 5 to 8 between observations and model estimates of intensity sensitivity to SST increases. Mann and Emanuel’s (2006) analysis of Atlantic basin data extending back to the late 19th century indicated a pronounced increase of TC frequency in the Atlantic since that time. This reported increase of TC frequency has not as yet been reconciled with existing theory or modeling of storm frequency. As noted previously, Landsea et al. (2004) maintain that TC counts for the Atlantic basin were likely undercounted in the earlier parts of the record, with the number of “missed” Atlantic basin tropical storms and hurricanes per year estimated to be on the order of 0-6 for the period 1851-85 and 0-4 for the period 1886-1910. 4.2.6. Simulations of future TC behavior a. Global and regional nested models Future changes in tropical storms projected by global or regional climate models (RCMs) are subject to many sources of uncertainty including: the future climate forcing scenario; initial conditions; regional patterns and magnitudes of future climate change for various fields; model physics and dynamics; and so forth. Since tropical storms are relatively rare events, large samples sizes (i.e. many years) are typically required to test the significance of any changes against natural variability, depending upon the

476

metric being examined. The changes in frequency of storms simulated by models are often smaller than the climatological bias of the models. These errors in the tropical storm climatology add to the uncertainty of the future changes in tropical storms projected by the models. The combined effect of all the sources of uncertainty is that there is large overall uncertainty in future changes in tropical cyclone frequency as projected by climate models forced with future greenhouse gases. The IPCC Third Assessment Report concluded that the TC frequency results of GCM experiments are inconclusive (see Giorgi et al. 2001). The most recent results obtained from medium and high resolution (120km-20km) GCMs (Table 1) indicate a consistent signal of fewer tropical cyclones globally in a warmer climate, though this finding is still not conclusive. While, these models consistently show a global decrease in frequency (e.g. Sugi et al. 2002; McDonald et al. 2005; Bengtsson et al. 2006; Oouchi et al. 2006), there are regional variations in the sign of the changes, and these vary substantially between models (Table 1). For example, more storms are projected in the North Atlantic region in some models, despite a large reduction globally (Sugi et al. 2002; Oouchi et al. 2006), while fewer Atlantic TCs are projected by the N144 HadAM3 atmosphere only model (McDonald et al. 2005). Chauvin et al. (2006) found that the sign of the changes in tropical cyclone frequency in the north Atlantic basin depended on the SST anomaly pattern in their stretched grid global model experiments (50km over Atlantic region). Walsh et al. (2004) using a 30 km grid nested regional model for the Australia region, found little change in frequency of tropical cyclones near Australia in their 3xCO2 RCM experiments. All of these results should be treated with caution, as it is not always clear that these changes are greater than the model’s natural variability, or that natural variability or the TC genesis process is properly simulated in the models. Concerning future changes in TC intensity, there is substantial disagreement among recent global and regional modeling studies, although the highest resolution models available show evidence for some increase of intensity. As discussed earlier, simulated future changes of intensity in current global models may not be reliable, since these models do not simulate the very intense TCs observed in the present climate, even in the case of the relatively high resolution (20km grid) simulation of Oouchi et al. (2006). Given this caveat, Tsutsui (2002), Walsh et al. (2004), McDonald et al. (2005) and Oouchi et al. (2006) all report evidence for intensity increases, while Sugi et al. (2002), Bengtsson et al. (2006), and Hasegawa and Emori (2005; western North Pacific only) , and Chauvin et al. (2006; North Atlantic only) found either no increase or a decrease of intensity. Among these studies, Tsutsui and Bengtsson et al. used relatively low resolution models; McDonald et al., Sugi et al., and Hasegawa and Emori used medium (~120km) grid spacing models; and Oouchi et al. and Walsh et al used relatively high resolution models. The Oouchi et al. (2006) study reports that the number of the most intense cyclones increases globally in their 20 km grid warming climate simulation, despite a large decrease in overall TC numbers. However, statistically significant intensity increases in their study were confined to only one or two individual basins. Walsh et al. (2004), focussing on the Australia region with a nested regional model, found little change in overall TC frequency under 3xCO2 conditions, but a 56% increase in the number of storms with relatively high maximum winds (>30 m/sec in their model), and a 26% increase in the number of storms with central pressures less than 970mb. Similarly, Knutson et al. (1998) simulated a significant CO2 warming-induced increase of typhoon intensities in the NW Pacific basin, based on downscaling a sample of 51 tropical storms from a high CO2 scenario of a global climate model into a regional nested hurricane model. Regarding TC precipitation based on global models, Hasegawa and Emori (2005) found an increase in TC-related precipitation in the western North Pacific, despite a decrease in TC intensity in their model. Chauvin et al (2006) found a similar result in the North Atlantic in their model. Yoshimura et al. (2006) found a similar result on a global domain. Hasegawa and Emori interpret the increase in hurricane-related precipitation as being due to enhanced atmospheric moisture in the warmer climate--a mechanism which has been discussed in the context of extreme precipitation in general by Trenberth (1999), Allen and Ingram (2002), and Emori and Brown (2005). Note that enhanced latent heating associated with increased TC precipitation does not necessarily lead to intensification of the TC,

477

since the enhanced heating is balanced to some degree by enhanced adiabatic cooling for given updraft due to the increased dry static stability in the simulated warmer climate (e.g., Sugi et al. 2002). Concerning observed changes in hurricane-related precipitation, Groisman et al. (2004) report finding no increasing trend in the total seasonal hurricane-related precipitation along the U.S. Southeast coast, despite finding that the frequency of very heavy precipitation unrelated to TCs has increased significantly in the same region and over the conterminous U.S. during the 20th century. They have not yet examined the behavior of hurricane-related precipitation on a per storm basis, and thus the time series they examine are influenced by changes in U.S. TC activity which has exhibited substantial multidecadal variability but no trend (Goldenberg et al. 2001; Landsea 2005). An important issue is to identify the underlying mechanisms producing changes in future TC behavior in the GCMs simulation. Sugi et al. (2002) report that the simulated reduction of global TC frequency in their model was closely related to the weakening of tropical circulation, which in turn resulted from a considerable increase in the dry static stability, coupled with relatively little increase in the precipitation. Yoshimura and Sugi (2005) performed a series of experiments to test the relative effects of SST changes and changes in CO2 on changes in TC frequency in their model. They found that the decrease in tropical storm frequency in their model was due to the increased CO2 (see also Sugi and Yoshimura 2004), with the SST changes having a relatively small direct impact. Regarding the regional variations in projected TC frequencies, the results of Sugi et al (2002) and McDonald et al (2005) and Chauvin et al. (2006) suggest that dynamical factors such as low level vorticity and vertical wind shear play a more important role than thermodynamical factors such as SST and moist instability. Knutson and Tuleya (2004) examined the mean vertical wind shear of the zonal wind for the tropical Atlantic basin in different coupled models. Their analysis showed a slight preference for increased vertical shear under high CO2 conditions if all of the models are considered, and a somewhat greater preference for increased shear if the three models with the poorest present-day simulation of shear in the basin are excluded. Examination of the trends in deep convection in the tropics provides some guidance on potential changes in tropical cyclogenesis regions. Dutton et al. (2000) examine changes in tropical convection in a fully coupled, transient CO2 simulation using the NCAR CSM1. They find that the SST threshold for tropical deep convection is about 25°C in the 1×CO2 climate, consistent with observational studies. This SST threshold increases as the level of CO2 and the global mean surface temperature increase in the model, to 26.0°C for 2×CO2 and 26.7°C for 3.4×CO2 (at the end of the 134 year simulation). Throughout this simulation, the area of convection between 40°N and 40°S remains approximately constant, however the precipitation intensity increases ~2%. b. Theoretical or idealized modeling frameworks Thus far, almost all theoretical or idealized modeling frameworks have focused on potential future changes in the intensities of TCs. Emanuel (1987) and later Tonkin et al. (1997) first presented evidence, based on potential intensity theory, that CO2-induced climate change as simulated by several GCMs implied significant increases in future TC intensities. Limitations of their approach were discussed in Henderson-Sellers et al. (1998). These theory-based assessments received model-based support from Knutson and Tuleya (2004), who used an idealized hurricane model framework to evaluate tropical climate warming scenarios from nine different coupled climate models, all forced by increasing CO2 levels. They reported a tropical cyclone intensity increase of about 3.3% per degree Celsius SST increase, which was roughly comparable to the increase predicted by the Emanuel and Holland potential intensity theories for those environments. The above methods account for changes in atmospheric temperature above the warming sea surface—an effect which acts to limit the increase of intensity for a given SST increase compared to the rate in the absence of the atmospheric temperature increases (e.g., Shen et al. 2000). Knutson et al. (2001) found that the CO2-warming-induced intensification of tropical cyclones in their idealized model

478

framework was robust to the inclusion of ocean coupling beneath the storms. Knutson and Tuleya (2004) reported that near-storm (100km radius) rainfall increased by about 10% per degree Celsius in their experiments. Wu and Wang (2004) performed an initial assessment of the potential impact of greenhouse gas-induced climate change on TC tracks using a trajectory modeling approach for the NW Pacific basin. Based on experiments derived from a particular climate model, they found evidence for inferred track changes, although the pattern of changes was fairly complex, as opposed to a more simply described, systematic change. Royer (1998) illustrated the use of a modified genesis parameter, based on a measure of convective rainfall as opposed to SST or oceanic heat content, and showed that TC frequency results for a future climate scenario depended strongly on whether the modified or unmodified genesis parameter approach was used. The empirical genesis index developed by Emanuel and Nolan (2004) and Nolan et al. (2006) implies a positive relation between potential intensity and the likelihood of tropical cyclogenesis. This finding, coupled to the increased potential intensity as inferred from increased CO2 climate model simulations (e.g., Emanuel 1987; Tonkin et al. 1997; Knutson and Tuleya 2004) implies a possible increase in tropical storm frequency in a warmer climate, unless other factors in their index (e.g., wind shear, vorticity, or relative humidity) change in ways to offset the impact of the potential intensity increase. In further recent work aimed at increasing the realism of simulated TC genesis, Nolan et al. (2006) have undertaken a very high-resolution (4 km grid) idealized modeling approach, using the Weather Research and Forecast Model (WRF) to explore the relationship between local values of potential intensity, the Coriolis parameter, and the likelihood of tropical cyclogenesis. Their initial results show that, in radiative-convective equilibrium (RCE), the potential for TC genesis increases with increasing values of SST. As the environmental surface wind is increased from zero, the genesis potential increases at first due to a substantial increase in the mid-level humidity, even though PI decreases. As the mean surface winds increase beyond 3 m s-1, both the PI and the genesis potential rapidly decline due to increased warming aloft by stronger and wetter convection. Additionally, Nolan et al. observed “spontaneous” TC genesis from random convection in RCE, suggesting that in very ideal environments, the absence of significant precursors such as easterly waves may not be a limiting factor on TC genesis. 4.2.7. The role of TCs in the climate system The possibility that tropical cyclones play an active as opposed to essentially passive role in the climate system was proposed by Emanuel (2001). According to this hypothesis, tropical cyclones, through wind-induced mixing of tropical ocean waters and subsequent re-heating of the cold wakes, make a potentially important contribution to the meridional heat transport by the oceans. Boos et al. (2004) provide additional support for this idea through idealized ocean modeling experiments. If in a warming climate, increased tropical storm activity substantially increases the poleward heat transport by the ocean through this mechanism, this process may then help explain the occurrence of distant past climates, such as the Eocene, characterized by strongly reduced equator-to-pole temperature gradients. With enhanced poleward oceanic heat transport, the high latitudes would warm more than otherwise, while the warming in tropical latitudes would be moderated. This in turn would moderate any projected increases in tropical cyclone intensity relative to those predicted on the basis of current global climate model simulations of future climates. In another example of the possible role of TC activity on climate, Hart (2006) has explored the impact of recurving tropical cyclone activity on the subsequent winter climate. He demonstrates that, for years with anomalously high numbers of recurving tropical cyclones in the Northern Hemisphere, the baroclinicity of the subsequent winter season is substantially reduced. It is hypothesized that this reduction in hemispheric baroclinicity is tied to snow cover (Hart et al. 2006).

479

4.2.8. Roadblocks to further advancements There are substantial roadblocks both in making reliable future projections about TC activity and in determining whether a trend can be detected in historical TC data. a. Data homogeneity in TC databases. For the climate change detection problem, a large hurdle is the quality of the tropical cyclone historical databases. The databases were populated over time without a focus on maintaining data homogeneity, a key requirement for databases which are to be used to assess possible climate-related trends. In some cases, such as the NW Pacific basin, our ability to monitor TC intensity has diminished over time. For example, aircraft reconnaissance was conducted in the NW Pacific basin beginning in the 1940s, but was discontinued in 1987. Experience with reanalysis of the HURDAT database for the Atlantic basin (Landsea et al. 2004) indicates that considerable effort and analysis is required to identify and attempt to correct, where possible, past problems with the TC databases. Indeed, even in 2006, operational satellite-based estimates of the intensity of TS Chris were found to be off by a full storm category when reconnaissance aircraft surveyed the storm. The possibility of such errors across all of the other ocean basins is real and problematic from both operational and climate perspectives. While reanalysis may help provide a more uniform assessment based upon consistent use of pressure-wind relationships and flight level to surface wind analyses, it will not recover hurricanes that were never observed. For example, over the open oceans before that advent of satellite coverage in the 1960s, there will never be a complete, reliable TC dataset for any of the basins. Even in the Atlantic, aircraft reconnaissance typically monitors about half of the tropical cyclones. However, it may be possible to have a high quality, global analysis of TC intensities and frequencies from about 1970 onward with substantial effort. One method that may be able to provide longer, reliable records is to focus upon analyses of landfalling tropical cyclones that have occurred along coastal regions with substantial populations. The tradeoff with this approach to get longer time series is that one only samples a much smaller number of tropical cyclones compared to the entire basin. The widespread concerns about problems in the TC databases reduces confidence in trends derived from those databases, and thus is an important roadblock to further advancement on the topic of historical TC trends. b. Data homogeneity concerns with other TC-related climate variables Improved understanding of the causes of past variations or trends in TC activity will depend on the existence of reliable climate-quality data sets for related variables, such as SST, atmospheric temperature, moisture, wind shear, etc. Although reanalysis efforts by NCEP/NCAR and ECMWF have led to important improvements in this regard, recent studies of upper-air data sets (e.g., Santer et al. 2005) identify likely remaining problems that could substantially affect efforts to reconcile historical TC behavior with various environmental influences. These data quality issues therefore also remain as an important roadblock for further advancement. [In using global reanalysis datasets such as NCAR/NCEP and ECMWF for TC-related studies (e.g., Sriver and Huber 2006), inclusion of new observations over time complicates monitoring of trends of tropical cyclone statistics, as improved observations should lead to better identification of tropical cyclones. While the circulation of larger tropical cyclones can be identified on the synoptic scale, some systems remain smaller scale (mesoscale) and the region in which intensity is defined (the maximum sustained surface winds in the eyewall) is always on the mesoscale, which implies that these features typically cannot be well-represented in low resolution reanalysis products.]

480

c. Limitations of climate models Climate models contain hypotheses for how the climate system works in a framework which allows experiments to be performed to test various hypotheses or compare the model’s historical simulations against historical observations. Nonetheless there are important uncertainties in climate models and the radiative forcings used for such experiments. For example, past aerosol forcing due to the interaction of aerosols with cloud and precipitation processes (indirect aerosol effects) remain quite uncertain. Many inferences about relative contributions of internal climate variability to past observed climate fluctuations rely on climate model simulations of internal variability, although paleo reconstructions provide important contributions to this question. Climate models have known limitations in simulating important internal modes of variability of the climate system (such as ENSO), although more recent models are improving in that regard (e.g., Wittenberg et al. 2006). The climate sensitivity to past and future radiative forcing is another important area of uncertainty, both on the global scale and with respect to important regional details, as evidenced by the wide range of likely global climate sensitivity to CO2 doubling (1.5-4.5o C) reported in the IPCC 3rd Assessment Report (IPCC 2001). In addition to climate sensitivity, there is considerable uncertainty in projections of future warming due to uncertainties in the rate of future ocean heat uptake as well as uncertainties in various climate forcing agents (IPCC 2001), including but not limited to greenhouse gas concentrations. These uncertainties combined lead to a wide range (1.4-5.8oC) in projected global warming by 2100 according to the IPCC (2001). Although the projected warming of tropical SSTs is generally smaller than the global mean warming, the above range provides an indication of the relative degree of uncertainty that also applies to future projections of tropical storm basin SSTs when forcing uncertainties are considered. The forcing-related uncertainty has not yet been formally assessed in detail at the tropical storm basin scale. The limited resolution of global climate models implies that many aspects of TC-like storms as simulated by the current models will not be very realistic, including the intensity and important smaller-scale structure such as the eye and eye-wall. This situation will gradually improve as available computing power increases (e.g., Oouchi et al. 2006). Meanwhile, questions remain about the realism of the TC genesis process in the global models. Generally, atmosphere-only models have been used for the global model-based TC-change assessments, as available computing power has been used for increasing atmospheric resolution rather than addition of ocean coupling. Eventually, this simplification will need to be relaxed, particularly in order to explore impacts of ocean coupling on TC genesis and intensity, as well as the possible role of TCs on the climate system (Section 7). The important impact of model physics and physical parameterizations, even in high-resolution models, means that future progress will depend on both increased scientific understanding and increased computing power. d. Limitations of high-resolution idealized models and theory While high-resolution idealized models can address the problem of limited resolution to some degree, this approach has limitations and uncertainties to be addressed. For example, the nested model used for the down-scaling may have a substantially different climatology and climate sensitivity from the “parent” global model, raising questions about the effect of such model incompatibilities on the reliability of the overall results obtained. The simulations can also be affected by the chosen domain (e.g., Landman et al. 2005). Clearly a preferred approach would be to avoid the downscaling issue altogether by using the TC statistics from the original GCM. The potential feedback of the TC activity on the climate system (section 7) also cannot be modeled using the simple one-way nesting approaches employed to date in TC/climate studies, nor can it be reliably inferred from the present generation of global models due to resolution limitations.

481

In contrast to the theory of potential intensity of TCs, which is more well-established, a comparable theory of TC frequency is not well-developed at this time. (We note that even the current theories of potential TC intensity include many assumptions and for example do not consider any dynamical limitations on TC intensity.) The lack of theoretical underpinning of TC genesis and frequency of occurrence remains as an important roadblock to progress in this area, apart from global model limitations. 4.2.9. Proposals for moving forward In general, hurricane-climate research is expected to progress most rapidly when a combination of theory, modeling, and observations are brought to bear on the problem. a. Improved paleoclimate, historical, and future TC databases The need for improved climate-quality tropical cyclone databases seems clear. These will provide better information for assessing future changes, and more reliable statistical assessments of past changes in hurricane activity, including land fall, in all basins. Specific examples include the need to reanalyze historical tropical cyclone databases in all basins, and not just the Atlantic. Such efforts should be encouraged and supported. Greater efforts should be made to provide researchers with access to original “raw” historical observations (i.e., ship, station, buoy, radar, aircraft, and satellite data) rather than derived quantities, concerning past tropical cyclones. Concerning future measurement systems, we advocate a comprehensive analysis of the optimal mix of observing systems in support of tropical cyclone measurement (for climate, forecasting, and other needs). Such an analysis should include consideration of both the overall costs and benefits of different mixes of observing platforms, with researchers and forecasters providing hard data on the benefits that a given mix of platforms can provide. As an example, aircraft reconnaissance was conducted in the NW Pacific basin beginning in the 1940s, but was discontinued in 1987 in favor of satellite-only intensity estimation. Is a resumption or initiation of manned or unmanned aircraft reconnaissance in various basins now justifiable in terms of costs, benefits, and alternative measurement techniques? A related issue is that future improvements in observing systems will lead, unfortunately, to more discontinuities and biases unless recognized and corrected for. For example, in 2007 the U.S. Air Force reconnaissance aircraft will be outfitted with Stepped Frequency Microwave Radiometers to more provide continuous surface wind estimates for the first time (Uhlhorn and Black 2003). Researchers need to be cognizant that large monitoring changes have occurred in the past and will continue to occur in the future, which can make determining actual climate-related trends problematic. Studies of how sampling can alter monitoring of both frequency and intensity of tropical cyclones are one approach to investigating the data homogeneity issue. For example, what would the 2005 Atlantic hurricane season look like using only the monitoring capabilities available in 1970, 1950, or 1900? Until better quantitative estimates of how the current observational network influences the determination of numbers and intensities of tropical cyclones, climate trends may be difficult to distinguish from changes induced by monitoring improvements. Paleotempestology research, which attempts to use information in the geological record to infer pre-historic hurricane activity, should continue to receive support from funding agencies. As the techniques themselves mature, thought should be given to a transition from technique-development research to systematic surveys designed to produce a comprehensive long-term record of tropical cyclone climatology.

482

b. Improved TC modeling Tropical cyclone/climate modeling studies will benefit from efforts to improve global climate modeling in general. In addition, studies which focus on simulation or downscaling of TCs could benefit from more rigorous testing of model performance with simulating a wider range of TC metrics. For example, the ability of models to simulate known interannual or interdecadal TC variability characteristics identified in various basins (e.g., Bell and Chelliah, 2006; Chan and Liu 2004) could be further exploited. A similar recommendation would also apply to studies using empirical approaches such as seasonal genesis parameters (see below). In TC climate change experiments with climate models, statistical significance testing should be emphasized to ensure that reported changes are not simply due to limited sampling. This may be particularly important in basins such as the Atlantic which feature large multi-decadal variations in some observed TC metrics. By analyzing several models using a common tropical storm metric, perhaps with common adjustments for resolution effects (e.g., Walsh et al. 2006), intercomparisons of sensitivity results between different models will be more informative. Such a procedure would help reduce differences between model results arising from differing analysis techniques alone. Analysis of individual models with perturbed physics experiments can be useful in isolating mechanisms producing model behavior. In general, there is a need to improve understanding of the physical mechanisms producing the climate-induced changes in TC behavior in the models. c. Improved empirical approaches to TC activity Exploration of empirical approaches, such as seasonal genesis parameters, should be encouraged, including testing/evaluation and improvements aimed at reproducing characteristics of historical TC activity in different basins from both observations and climate model simulations. Based on these results, these approaches may be useful for making climate change projections of TC activity, although caution must be exercised (e.g. Ryan et al. 1992). A similar recommendation would apply to studies leading to the development of empirical approaches for tropical cyclone potential intensity. Such empirical approaches should include not only thermodynamic parameters, such as the SST and outflow temperature, but also the environmental dynamical parameters that control TC intensity, such as the vertical wind shear and translational speed (Zeng et al. 2006). Current global climate models can simulate the large-scale circulation much more realistically than the individual TCs. Thus empirical approaches with environmental parameters as input to estimate TC potential intensity should be further exploited in this area. Bibliography Allen, M. and W. Ingram, 2002: Constraints on future changes in climate and the hydrological cycle, Nature, 419, 224-232.

Barnett, T. P., D.W. Pierce, K. M. AchutaRao, P. J. Gleckler, B. D. Santer, J. M. Gregory, and W. M. Washington, 2006: Penetration of human-induced warming into the world’s oceans. Science, 309, 284-287.

Bell, G. D., and M. Chelliah, 2006: Leading tropical modes associated with interannual and multidecadal fluctuations in North Atlantic hurricane activity. J. Climate, 19, 590-612.

Bengtsson, L., M. Botzet, and M. Esch, 1995: Hurricane-type vortices in a general circulation model. Tellus 47A: 175-196

Bengtsson, L., K. Hodges, and E. Roeckner, 2006: Storm tracks and climate change. J. Climate, 19,

483

3518-3543.

Boos, W. R., J. R. Scott and K. A. Emanuel, 2004: Transient diapycnal mixing and the meridional overturning circulation. J. Phys. Ocean., 34, 334-341

Camargo, S., A. G. Barnston, and S. E. Zebiak, 2005: A statistical assessment of tropical cyclone activity in atmospheric general circulation models. Tellus 57A: 589-604, DOI: 10.1111/j.1600-0870.2005.00117.

Camargo, S., K. A. Emanuel, and A. H. Sobel, 2006: ENSO and genesis potential index in reanalysis and AGCMs. Proc. 27th AMS Conf. on Hurr. and Trop. Meteor., abstract #15C.2. http://ams.confex.com/ams/pdfpapers/108038.pdf

Camp, J. P., and M. T. Montgomery, 2001: Hurricane maximum intensity: past and present. Mon. Wea. Rev., 129, 1704-1717.

Chan, J.C. L., 1985: Tropical cyclone activity in the northwest Pacific in relation to the El Niño / Southern Oscillation phenomenon. Mon. Wea. Rev., 113: 599-606

Chan, J. C. L., 2006: Comment on “Changes in Tropical Cyclone Number, Duration, and Intensity in a Warming Environment, Science, v. 311, p. 1713.

Chan, J. C. L., and K. S. Liu, 2004: Global warming and Western North Pacific typhoon activity from an observational perspective. J. Climate, 17, 4590-4602.

Chan, S., and J. L. Evans, 2002: Comparison of the structure of the ITCZ in the West Pacific during the boreal summers of 1989-1993 using AMIP simulations and ECMWF reanalysis. J. Climate, 15, 3549-3568.

Chauvin, F., J.-F. Royer and M. Déqué, 2006: Response of hurricane-type vortices to global warming as simulated by ARPEGE-Climat at high resolution. Clim. Dyn., 27, 377-399

Chenoweth, M. 2003. The 18th century climate of Jamaica. American Philosophical Society.

Delworth, T. L., and T. R. Knutson, 2000: Simulation of early 20th Century global warming. Science, 287(5461), 2246-2250.

Delworth, T. L., and M. E. Mann, 2000: Observed and simulated multidecadal variability in the Northern Hemisphere. Clim. Dyn., 16(9), 661-676.

DeMott, C. A., and D. A. Randall, 2004: Observed variations of tropical convective available potential energy. J. Geophys. Res., 109, D02102, doi:10.1029/2003JD003784.

Donnelly, J. P., S. S. Bryant, J. Butler, J. Dowling, L. Fan, N. Hausmann, P. Newby, B. Shuman, J. Stern, K. Westover, and T. Webb, III, 2001a: A 700 yr sedimentary record of intense hurricane landfalls in southern New England. Geol. Soc. Amer. Bull., 113, 714-727.

Donnelly, J. P., S. Roll, M. Wengren, J. Butler, R. Lederer, and T. Webb, III, 2001b: Sedimentary evidence of intense hurricane strikes from New Jersey. Geology, 29, 615-618

Donnelly, J. P. and T. Webb, III,. 2004: Backbarrier sedimentary records of intense hurricane landfalls in the northeastern United States. In: Hurricanes and Typhoons: Past, Present, and Future (eds. R.J. Murnane, R. J. and K-b. Liu), p. 58-95. Columbia University Press.

484

Donnelly, J.P., J. Butler, S. Roll, M. Wengren, and T. Webb, III, 2004. A backbarrier overwash record of intense storms from Brigantine, New Jersey. Marine Geology, 210, 107-121

Donnelly, J P. 2005. Evidence of past intense tropical cyclones from backbarrier salt pond sediments: a case study from Isla de Culebrita, Puerto Rico, USA. J. Coastal Res., 42, 201-210.

Dutton, J. F. , C. J. Poulsen, and J. L. Evans, 2000: The effect of global climate change on the regions of tropical convection in CSM1. Geophysical Research Letters, 27, 3049.

Elsner, J. B., 2006: Evidence in support of the climate change-Atlantic hurricane hypothesis. Geophys. Res. Lett., 33, L16705, doi:10.1029/2006GL026869.

Elsner, J. B., K-b. Liu, and B. L. Kocher, 2000: Spatial variations in major U.S. hurricane activity: Statistics and a physical mechanism. J. Climate, 13, 2293-2305.

Elsner, J. B., R. J. Murnane, and T. H. Jagger, 2006: Forecasting U.S. hurricanes 6 months in advance. Geophys. Res. Lett., accepted for publication. Available online at: http://garnet.acns.fsu.edu/~jelsner/PDF/Research/LongLead.pdf

Elsner, J. B., A. A. Tsonis, and T. H. Jagger, 2006: High-frequency variability in hurricane power dissipation and its relationship to global temperature. Bull. Amer. Meteor. Soc., 87, 763-768.

Emanuel, K.A., 1987: The dependence of hurricane intensity on climate. Nature, 326, 483-485.

Emanuel, K. A., 1999: Thermodynamic control of hurricane intensity. Nature, 401, 665-669.

Emanuel, K., 2000: A statistical analysis of tropical cyclone intensity. Mon. Wea. Rev., 128, 1139-1152.

Emanuel, K.A., 2001: The contribution of tropical cyclones to the oceans’ meridional heat transport. J. Geophys. Res., 106, D14, 14,771-14,781.

Emanuel, K. A., 2005a: Increasing destructiveness of tropical cyclones over the past 30 years. Nature, 436, 686-688.

Emanuel, K. A., 2005b: Emanuel replies. Nature, 438, doi:10.1038/nature04427.

Emanuel, K., 2006: Environmental influences on tropical cyclone variability and trends. Proc. 27th AMS Conf. on Hurr. and Trop. Meteor., #4.2. Available online at: http://ams.confex.com/ams/pdfpapers/107575.pdf

Emanuel, K., S. Ravela, E. Vivant, and C. Risi. 2006: A statistical deterministic approach to hurricane risk assessment. Bull. Amer. Meteor. Soc.,: 87, 299–314.

Emanuel, K. and D. S. Nolan, 2004: Tropical cyclone activity and the global climate system. Proc. 26th AMS Conf. on Hurr. and Trop. Meteor., #10A.2, p. 240-241.

Emori, S., A. Hasegawa, T. Suzuki and K. Dairaku, 2005: Validation, parameterization dependence, and future projection of daily precipitation simulated with a high-resolution atmospheric GCM. Geophys. Res. Lett., 32, L06708, doi:10.1029/2004GL022306.

Emori, S. and S. J. Brown, 2005: Dynamic and thermodynamic changes in mean and extreme precipitation under changed climate. Geophys. Res. Lett., 32, L17706, doi:10.1029/2005GL023272.

485

Enfield, D.B., and A.M. Mestas-Nuñez, 2000: Global modes of ENSO and non-ENSO SST variability and their associations with climate. In: El Niño and the Southern Oscillation: Multiscale Variability and Global and Regional Impacts, H.F. Diaz and V. Markgraf, Eds., Cambridge University Press (ISBN 0521621380), p. 89-112.

Fernandez-Partagas, J. and H. F. Diaz, 1996: Atlantic hurricanes in the second half of the nineteenth century. Bull. Amer. Meteor. Soc., 77, 2899-2906.

Frappier, A.B., Sahagian, D., Carpenter, S.J., Gonzalez, L.A. and Frappier, B.R., 2006: A stalagmite stable isotope record of recent tropical cyclone events. Geology, (in review).

Free, M., M. Bister, and K. Emanuel, 2004: Potential intensity of tropical cyclones: comparison of results from radiosonde and reanalysis data. J. Climate, 17, 1722-1727.

Garcia Herrera, R., L. Gimeno, P. Ribera, and E. Hernandez, 2005. New records of Atlantic hurricanes from Spanish documentary sources. J. Geophys. Res., 110: D03109.

Garcia Herrera, R., F. Rubio, D. Wheeler, E. Hernandez, M. R. Prieto, and L. Gimero, 2004: The use of Spanish and British documentary sources in the investigation of Atlantic hurricane incidence in historical times. In: Hurricanes and Typhoons: Past, Present, and Future (eds. R.J. Murnane, R. J. and K-b. Liu), p. 149-176. Columbia University Press.

Gettelman, A. D. J. Seidel, M. C. Wheeler, and R. J. Ross, 2002: Multidecadal trends in tropical convective available potential energy. J. Geophys. Res., 107, 4606, doi:10.1029/2001JD001082.

Giorgi, F., B. Hewitson, J. Christensen, M. Hulme, H. Von Storch, P. Whetton, R. Jones, L. Mearns, and C. Fu, 2001: Regional climate information - evaluation and projections. In Climate Change 2001: The Scientific Basis. Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change. [Houghton JT, Ding Y, Griggs DJ, Noguer M, van der Linden PJ, Dai X, Maskell K and Johnson CA, eds.] 881 pp. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA.

Goldenberg, S. B., C. W. Landsea, A.M. Mesta-Nuñez, and W. M. Gray, 2001: The recent increase in Atlantic hurricane activity: causes and implications. Science, 293, 474-479.

Gray, W. M., 1979: Hurricanes: their formation, structure, and likely role in the tropical circulation. In Meteorology over the tropical oceans. Ed., D. B. Shaw, Roy. Meteor. Soc., 155-218.

Gray, W. M., 1990: Strong association between West African rainfall and U.S. landfall of intense hurricanes. Science, 249, 1251-1256.

Hart, R. E., 2006: The winter impact of recurving tropical cyclones. 27th AMS Conference on Hurricanes and Tropical Meteorology, 24-28 April 2006, Monterey, CA.

Hart, R. E., L. F. Bosart, and C. Hosler, 2006: The possible hemispheric impacts of anomalous recurving tropical cyclone frequency. Submitted to Mon. Wea. Rev., August 2006.

Hasegawa, A. and S. Emori, 2005: Tropical cyclones and associated precipitation over the Western North Pacific: T106 atmospheric GCM simulation for present-day and doubled CO2 climates. SOLA, 1, 145-148, SOI:10.2151/sola.2005-038

Hayne, M. and J. Chappell. 2001. Cyclone frequency during the last 5,000 years from Curacoa Island, Queensland. Palaeogeography, Palaeoclimatology, Palaeoecology, 168, 201-219.

486

Held, I. M., T. L. Delworth, J. Lu, K. L. Findell, and T. R. Knutson, 2005: Simulation of Sahel drought in the 20th and 21st centuries. Proc. Nat. Acad. Sci., 102(50), 17891-17896.

Henderson-Sellers, A., H. Zhang, G. Berz, K. Emanuel, W. Gray, C. Landsea, G. Holland, J. Lighthill, S-L. Shieh, P. Webster, and K. McGuffie, 1998: Tropical cyclones and global climate change: a post-IPCC assessment. Bull. Amer. Meteor. Soc., 79, 19-38.

Holland, G.J., 1997: The maximum potential intensity of tropical cyclones. J. Atmos. Sci., 54, 2519-2541.

Hoskins B. J. and K. I. Hodges, 2005: A new perspective on southern hemisphere storm tracks. J. Climate, 18, 4108-4129

Hoyos, C. D., P. A. Agudelo, P. J. Webster, and J. A. Curry, 2006: Deconvolution of the factors contributing to the increase in global hurricane intensity. Science, 312: 94-97.

IPCC, 2001. Climate Change 2001: The Scientific Basis. Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change [Houghton, J. T., Y. Ding, D. J. Griggs, M. Noguer, P. J. van der Linden, X. Dai, K. Maskell, and C. A. Johnson (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, 881pp.

International Ad Hoc Detection and Attribution Group (IADAG), 2005: Detecting and attributing external influences on the climate system: a review of recent advances. J. Climate, 18, 1291-1314.

Kamahori, H. N. Yamazaki, N. Mannoji, and K. Takahashi, 2006: Variability in intense tropical cyclone days in the western North Pacific. SOLA, 2, 104-107, doi:10.2151/sola.2006-027.

Karoly, D. J., and Q. Wu, 2005: Detection of regional surface temperature trends. J. Climate, 18, 4337-4343.

Klotzbach, P. J., 2006: Trends in global tropical cyclone activity over the past twenty years (1986-2005), Geophys. Res. Lett., 33, L10805, DOI:10.1029/2006GL025881

Knaff, J. A, and C. R. Sampson, 2006: Reanalysis of West Pacific tropical cyclone intensity 1966-1987. Proceedings of 27th AMS Conference on Hurricanes and Tropical Meteorology, #5B.5. Available online at: http://ams.confex.com/ams/pdfpapers/108298.pdf

Knight, J. R., R. J. Allan, C. K. Folland, M. Vellinga, and M. E. Mann, 2005: A signature of persistent natural thermohaline circulation cycles in observed climate, Geophys. Res. Lett., 32, L20708, doi:10.1029/2005GL024233.

Knight, J. R., C. K. Folland, and A. A. Scaife, 2006: Climate impacts of the Atlantic Multidecadal Oscillation. Geophys. Res. Lett., 33, 17, L17706 http://dx.doi.org/10.1029/2006GL026242.

Knutson, T. R., T. L. Delworth, K. W. Dixon, I. M. Held, J. Lu, V. Ramaswamy, D. Schwarzkopf, G. Stenchikov, and R. J. Stouffer, 2006. Assessment of twentieth-century regional surface temperature trends using the GFDL CM2 coupled models. J. Climate, 19(9), 1624-1651.

Knutson, T. R., and R. E. Tuleya, 2004: Impact of CO2-induced warming on simulated hurricane intensity and precipitation: sensitivity to the choice of climate model and convective parameterization. J. Climate, 17, 3477-3495.

Knutson, T. R., and R. E. Tuleya, 2005: Reply. J. Climate, 18, 5183-5187.

487

Knutson, T. R., R. E. Tuleya, W. Shen, and I. Ginis, 2001: Impact of CO2-induced warming on hurricane intensities as simulated in a hurricane model with ocean coupling. J. Climate, 14, 2458-2468.

Knutson, T. R., R. E. Tuleya, and Y. Kurihara, 1998: Simulated increase of hurricane intensities in a CO2-warmed climate. Science, 279(5353), 1018-1020.

Landman, W. A., A. Seth, and S. J. Camargo, 2005: The effect of regional climate model domain choice on the simulation of tropical cyclone-like vortices in the Southwestern Indian Ocean. J. Climate, 18, 1253-1274, DOI: 10.1175/JCLI3324.1

Landsea, C. W., 1993: A climatology of intense (or major) Atlantic hurricanes. Mon. Wea. Rev., 121, 1703-1713.

Landsea, C. W., C. Anderson, N. Charles, G. Clark, J. Dunion, J. Fernandez-Partagas, P. Hungerford, C. Neumann, and M. Zimmer, 2004: The Atlantic hurricane database re-analysis project: Documentation for the 1851-1910 alterations and additions to the HURDAT database. In: Hurricanes and Typhoons: Past, Present and Future, R. J. Murname and K.-B. Liu, Eds., Columbia University Press, p. 177-221.

Landsea, C. W., 2005: Hurricanes and global warming. Nature, 438, doi:10.1038/nature04477.

Landsea,C.W., B.A.Harper, K.Hoarau, and J.A.Knaff, 2006: Can we detect trends in extreme tropical cyclones? Science. 313, 452-454.

Lawrence, J. R. and S. D. Gedzelman, 1996: Low stable isotope ratios of tropical cyclone rains. Geophys. Res. Lett., 23, 527-530.

Liu, K-b., 2004: Paleotempestology: Principles, Methods, and Examples from Gulf Coast Lake Sediments. In: Hurricanes and Typhoons: Past, Present, and Future, R.J. Murnane, R. J. and K-b. Liu, Eds., Columbia University Press, p. 13-57..

Liu, K-b. and M. L. Fearn, 1993: Lake-sediment record of late Holocene hurricane activities from coastal Alabama. Geology, 21, 793-796.

Liu, K-b. and M. L. Fearn, 2000: Reconstruction of prehistoric landfall frequencies of catastrophic hurricanes in northwestern Florida from lake sediment records. Quaternary Research, 54, 238-245.

Liu, K-b., C. Shen, C, and K. S. Louie, 2001: A 1000-year history of typhoon landfalls in Guangdong, southern China, reconstructed from Chinese historical documentary records. Annals of the Association of American Geographers 91, 453-464.

Louie, K.S. and K.-b. Liu, 2003: Earliest historical records of typhoons in China. J. Historical Geography, 29, 299-316.

Louie, K.S. and K.-b. Liu, 2004: Ancient records of typhoons in Chinese historical documents. In: Hurricanes and Typhoons: Past, Present, and Future , R.J. Murnane, R. J. and K-b. Liu, Eds., Columbia University Press, p. 222-248..

Ludlam, D.M., 1963: Early American hurricanes, 1492-1870. American Meteorological Society.

Mann, M. and K. Emanuel, 2006: Atlantic hurricane trends linked to climate change. EOS, 87, 233-241..

Marshall, G. J., P. A. Stott, J. Turner, W. M. Connolley, J. C. King, T. A. Lachlan-Cope, 2004: Causes of exceptional atmospheric circulation changes in the Southern Hemisphere. Geophys. Res. Lett.,

488

31(14), L14205,doi:10.1029/2004GL019952.

McDonald, R. E., D. G. Bleaken, D. R. Cresswell, V. D. Pope, and C. A. Senior, 2005: Tropical storms: representation and diagnosis in climate models and the impacts of climate change. Clim. Dyn., 25: 19-36, DOI: 10.1007/s00382-004-0491-0.

Mears, C. A., and F. J. Wentz, 2005: The effect of diurnal correction on satellite-derived lower tropospheric temperature. Science, 309, 1548-1551.

Meehl, G. M., W. M. Washingon, C. M. Amman, J. M. Arblaster, T. M. L. Wigley, and C. Tebaldi, 2004: Combinations of natural and anthropogenic forcings in twentieth-century climate. J. Climate, 17, 3721-3727.

Michaels, P. J., P. C. Knappenberger, and C. Landsea, 2005: Comments on “Impacts of CO2-induced warming on simulated hurricane intensity and precipitation: Sensitivity to the choice of climate model and convective scheme.” J. Climate, 18, 5179-5182.

Michaels, P. J., P. C. Knappenberger, and R. E. Davis, 2006: Sea-surface temperatures and troical cyclones in the Atlantic basin. Geophys. Res. Lett., 33, L09708, doi:10.1029/2006GL025757.

Millas, J.C., 1968: Hurricanes of the Caribbean and adjacent regions, 1492-1800. Academy of the Arts and Sciences of the Americas.

Miller, D. L., C. I. Mora, H. D. Grissino-Mayer, C. J. Mock, and M. E. Uhle, 2003: Tropical cyclone activity and climate fluctuations captured by oxygen isotopes in tree-ring cellulose from the Southeastern U.S. Abstracts, American Geophysical Union Fall 2003 Meeting.

Mock, C.J., 2004: Tropical cyclone reconstructions from documentary records: examples for South Carolina, United States. In: Hurricanes and Typhoons: Past, Present, and Future, R.J. Murnane, R. J. and K-b. Liu, Eds., Columbia University Press, p. 121-148.

Nolan, D. S., E. D. Rappin, and K. A. Emanuel, 2006: Could hurricanes form from random convection in a warmer world? Preprints, 27th AMS Meeting on Hurricanes and Tropical Meteorology, Monterey, CA. Available online at: http://ams.confex.com/ams/pdfpapers/107936.pdf

Nott, J. and M. Hayne. 2001. High frequency of “super-cyclones” along the Great Barrier Reef over the past 5,000 years. Nature, 413, 508-512.

Nguyen, K. C., and K. J. E. Walsh, 2001: Interannual, decadal, and transient greenhouse simulation of tropical cyclone-like vortices in a regional climate model of the South Pacific. J. Climate, 14, 3043-3054.

Oouchi, K., J.Yoshimura, H. Yoshimura, R. Mizuta, S. Kusunoki, and A. Noda, 2006: Tropical cyclone climatology in a global-warming climate as simulated in a 20km-mesh global atmospheric model: frequency and wind intensity analysis. J. Meteorol. Soc. Japan, , 84, 259-276.

Pezza, A. B., and I. Simmonds, 2005: The first South Atlantic hurricane: unprecedented blocking, low shear, and climate change. Geophys. Res. Lett., 32, L15712, doi:10.1029/2005GL023390.

Pielke, R. A., Jr., 2005: Are there trends in hurricane destruction? Nature, 438, E11

Rotstayn, L. and U. Lohmann, 2002: Tropical rainfall trends and the indirect aerosol effect. J. Climate, 15, 2103-2116.

489

Royer. J.-F., F. Chauvin, B. Timbal, P. Araspin, and D. Grimal, 1998: A GCM study of the impact of greenhouse gas increase on the frequency of occurrence of tropical cyclones. Clim. Change, 38, 307-343.

Ryan, B.F., I. G. Watterson, and J. L. Evans, 1992: Tropical cyclones frequencies inferred from Gray's yearly genesis parameter: validation of GCM tropical climate. Geophys. Res. Lett., 19, 1831-1834.

Santer, B. D., and Coauthors, 2005: Amplification of surface temperature trends and variability in the tropical atmosphere. Science, 309, 1551-1556.

Santer, B. D., and Coauthors, 2006: Forced and unforced ocean temperature changes in Atlantic and Pacific tropical cyclogenesis regions. Proc. Nat. Acad. Sci., 103, 13905-13910, 10.1073/pnas.0602861103..

Saunders, M. A., and A. S. Lea, 2005: Seasonal prediction of hurricane activity reaching the coast of the United States. Nature, 434, 1005-1008.

Shen, W., R.E. Tuleya, and I. Ginis, 2000: A sensitivity study of the thermodynamic environment on GFDL model hurricane intensity: implications for global warming. J. Climate, 13, 109-121.

Sherwood, S. C., J. R. Lanzante, and C. L. Meyer, 2005: Radiosonde daytime biases and late-20th century warming. Science, 309, 1556-1559.

Sriver, R. and M. Huber, 2006: Low frequency variability in globally integrated tropical cyclone power dissipation. Geophys. Res. Lett., 33, L11705, doi:10.1029/2006GL026167.

Stott, P. A., S. F. B. Tett, G. S. Jones, M. R. Allen, J. F. B. Mitchell, and G. J. Jenkins, 2000: External control of 20th century temperature by natural and anthropogenic forcings. Science, 290, 2133-2137.

Sugi, M., A. Noda, and N. Sato, 2002: Influence of global warming on tropical cyclone climatology: an experiment with the JMA global model. J. Meteorol. Soc. Japan 80: 249-272, DOI:10.2151/jmsj.80.249.

Sugi, M., and J. Yoshimura, 2004: A mechanism of tropical precipitation change due to CO2 increase. J. Climate, 17, 238-243.

Tang, B. H., and J. D. Neelin, 2004: ENSO influence on Atlantic hurricanes via tropospheric warming. Geophys. Res. Lett., 31, L24204, doi:10.1029/2004GL021072.

Tonkin, H., G. J. Holland, N. Holbrook, and A. Henderson-Sellers, 2000: An evaluation of thermodynamic estimates of climatological maximum potential tropical cyclone intensity. Mon. Wea. Rev., 128, 746-762.

Tonkin, H., G. Holland, C. Landsea, and S. Li, 1997: Tropical cyclones and climate change: a preliminary assessment. In Assessing Climate Change: Results from the Model Evaluation and Consortium for Climate Assessments, W. Howe and A. Henderson-Sellers, eds., Taylor and Francis Group, plc., London, 327-360.

Trenberth, K., 1999: Conceptual framework for changes of extremes of the hydrological cycle with climate change. Clim. Change, 42, 327-339.

Trenberth, K., 2005: Uncertainty in hurricanes and global warming. Science, 308, 1753-1754.

Trenberth, K. E., J. Fasullo, and L. Smith, 2005: Trends and variability in column-integrated atmospheric water vapor. Clim. Dyn., 24: 741-758.

490

Trenberth, K. E., and D. J. Shea, 2006: Atlantic hurricanes and natural variability in 2005. Geophys. Res. Lett., 33, L12704, doi:10.1029/2006GL026894.

Tsutsui, J., 2002: Implications of anthropogenic climate change for tropical cyclone activity: a case study with the NCAR CCM2. J. Meteorol. Soc. Japan, 80, 45-65, DOI:10.2151/jmsj.80.45

Tsutsui, J., and A. Kasahara, 1996: Simulated tropical cyclones using the National Center for Atmospheric Research community climate model. J. Geophys. Res. 101, 15013-15032.

Uhlhorn, E. W, and P. G. Black, 2003: Verification of remotely sensed sea surface winds in hurricanes. J. Atmos. and Ocean. Tech., 20, 99-116.

Vecchi, G. A., B. J. Soden, A. T. Wittenberg, I. M. Held, A. Leetmaa, and M. J. Harrison, 2006: Weakening of the tropical Pacific atmospheric circulation due to anthropogenic forcing. Nature, 441, 73-76. doi:10.1038/nature/04744.

Vitart, F., J. L. Anderson, and W. F. Stern, 1997: Simulation of inter-annual variability of tropical storm frequency in an ensemble of GCM integrations. J. Climate, 10, 745-760.

Vitart, F. and J. L. Anderson, 2001: Sensitivity of Atlantic tropical storm frequency to ENSO and interdecadal variability of SSTs in an ensemble of AGCM integrations. J. Climate, 14( 4), 533-545.

Vitart, F., J. L. Anderson, J. Sirutis, R. E. Tuleya, 2001: Sensitivity of tropical storms simulated by a general circulation model to changes in cumulus parametrization. Quart. J. Roy. Meteorol. Soc. 127, 25-51.

Walsh, K., 2004: Tropical cyclones and climate change: unresolved issues. Clim. Res., 27, 77-83.

Walsh, K. J. E., K.-C. Nguyen, and J. L. McGregor, 2004: Fine-resolution regional climate model simulations of the impact of climate change on tropical cyclones near Australia. Clim. Dyn., 22, 47-56, DOI: 10.1007/s00382-003-0362-0

Walsh, K. J. E. and B. F. Ryan, 2000: Tropical cyclone intensity increase near Australia as a result of climate change. J. Climate, 13, 3029-3036.

Walsh, K. J. E., M. Fiorino, C. W. Landsea, and K. McInnes, 2006: Objectively-determined resolution-dependent threshold criteria for the detection of tropical cyclones in climate models and reanalyses. J. Climate, (in press).

Watterson, I. G., J. L. Evans, and B. F. Ryan, 1995: Seasonal and inter-annual variability of tropical cyclogenesis: diagnostics from large-scale fields. J. Climate, 8, 3052-3066.

Webster, P. J., G. J. Holland, J. A. Curry, and H.-R. Chang, 2005: Changes in tropical cyclone number, duration, and intensity in a warming environment. Science, 309, 1844-1846.

Webster, P. J., J. A. Curry, J. Liu, and G. J. Holland, 2006: Response to comment on “Changes in tropical cyclone number, duration, and intensity in a warming environment”. Science, 311, 1713c.

Wittenberg, A. T., A. Rosati, N.-C. Lau, and J. J. Ploshay, 2006: GFDL's CM2 global coupled climate models - Part 3: Tropical Pacific climate and ENSO. J. Climate, 19(5), 698-722.

Wu, L., and B. Wang, 2004: Assessing impacts of global warming on tropical cyclone tracks. J. Climate, 17, 1686-1698.

491

Wu, L., B. Wang, and S. Geng (2005), Growing typhoon influence on east Asia, Geophys. Res. Lett., 32, L18703, doi:10.1029/2005GL022937.

Yoshimura, J. and M. Sugi, 2005: Tropical cyclone climatology in a high-resolution AGCM - impacts of SST warming and CO2 increase. SOLA, 1, 133-136, DOI: 10.2151/sola.2005-035.

Yoshimura, J., M. Sugi and A. Noda, 2006: Influence of greenhouse warming on tropical cyclone frequency. J. Meteor. Soc. Japan, 84, 405-428.

Zeng, Z., Y. Wang, and C.-C. Wu, 2006: Environmental dynamical control of tropical cyclone intensity-An observational study. Mon. Wea. Rev., (in press).

Zhang, R., and T. L. Delworth, 2006: Impact of Atlantic multidecadal oscillations on India/Sahel rainfall and Atlantic hurricanes. Geophys. Res. Lett., 33, L17712, doi:10.1029/2006GL026267.

492

Table 1. Summary of future changes in tropical storm frequency as simulated by climate GCMs under global warming conditions Reference Model Resolution Experiment Ratio (%) of number of storms in global warming experiment to number

in control experiment Global NH SH Ocean basin

N Atlantic

NW Pacific

NE Pacific

N Indian

S Indian

SW Pacific

Sugi et al. 2002 JMA timeslice

T106 L21 (~120km)

10y 1xCO2, 2xCO2

66 72 61 161 34 33 109 43 69

Tsutsui 2002 NCAR CCM2 T42 L18 10y 1xCO2 2xCO2 from 115y CO2 1% pa

102 86 111 91 116 124 99

McDonald et al. 2005

HadAM3 timeslice

N144 L30 (~100km)

15y IS95a 1979-1994 2082-2097

94 97 90 75 70 180 142 110 82

Hasegawa and Emori 2005

CCSR/NIES/FRCGC timeslice

T106 L56 (~120km)

5x20y at 1xCO2 7x20y at 2xCO2

96

Yoshimura et al. 2006

JMA timeslice

T106 L21 (~120km)

10y 1xCO2, 2xCO2

85

Bengtsson et al. 2006

ECHAM5-OMT63 L31 1.5° L40

A1B 3 members 30y 20C and 21C

94

Oouchi et al. 2006MRI/JMA timeslice

TL959 L60 (~20km)

10y A1B 1982-1993 2080-2099

70 72 68 134 62 66 48 72 57

Red = significantly more tropical storms in the future simulation Blue = significantly fewer tropical storms in the future simulation Black = not significant or significance level not tested

493

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES Topic 4.3 : Short-term Climate (Seasonal and Intra-seasonal) Prediction of Tropical Cyclone Activity and Intensity Rapporteur: Suzana J. Camargo

International Research Institute for Climate and Society Email : suzana@iri.columbia.edu Working Group: Maritza Ballester, Anthony G. Barnston, Phil Klotzbach, Paul Roundy, Mark A. Saunders, Frédéric Vitart, Matthew C. Wheeler

4.3.1 Seasonal Tropical Cyclone Forecasts 4.3.1a. Statistical Seasonal Tropical Cyclone Forecasts Seasonal tropical cyclone forecasts are currently produced using statistical and dynamical methods in various centers and for different regions. Statistical seasonal tropical cyclone prediction was first conducted in the Atlantic basin (Gray 1984a, 1984b) at Colorado State University using statistical relationships between Atlantic tropical cyclone activity and predictors such as the El Niño – Southern Oscillation (ENSO), the Quasi-Biennial Oscillation (QBO) and Caribbean basin sea level pressures. Statistical forecast techniques have continued to develop since these early forecasts (e.g. Gray et al. 1992, Klotzbach and Gray 2004). Additional groups since then began issuing statistical seasonal hurricane forecasts for the Atlantic including the Institute of Meteorology of Cuba (1996), NOAA (1998) and Tropical Storm Risk (1999). Currently, the Cuban seasonal forecast is based on the solution of a regression and an analogue method which gives various hurricane parameters, such as the total number of named storms, hurricanes and hurricane destruction potential for the entire Atlantic region, as well as separated numbers of named storms for the Caribbean Sea and the Gulf of Mexico, and the first and last day of the hurricane season (Ballester et al. 2004a, 2004b). The NOAA (National Oceanic and Atmospheric Administration) Atlantic Outlook is based on the state of the Atlantic multi-decadal signal (Goldenberg et al. 2001, Bell and Chelliah 2006) and the ENSO conditions. The NOAA outlook gives tercile probabilities for tropical cyclone activity level for various parameters (number of named storms, hurricanes, and major-hurricanes, and ACE [accumulated cyclone energy]). Since 2003 NOAA has also been issuing similar outlooks for the Eastern Pacific hurricane season, first experimentally, and operationally since 2005. Johnny Chan and colleagues at the City University of Hong Kong have issued seasonal forecasts for the Northwest Pacific basin (number of tropical cyclones, tropical storms and typhoons) since 1997 utilizing various environmental conditions, the most prominent ones being ENSO and the extent of the Pacific subtropical ridge (Chan et al. 1998, 2001, Liu and Chan, 2003). The Tropical Storm Risk (TSR) issues statistical forecasts for tropical cyclone activity in the Atlantic, western North Pacific and Australian regions. In the case of the western North Pacific, the seasonal predictability of ACE index has been computed (Lea and Saunders 2006). The seasonal prediction model uses Niño 3.75 (5°S-5°N, 180°-140°W) forecasts (Lloyd-Hughes et al. 2004) to predict the NW

494

Pacific ACE index. The NW Pacific ACE index is forecast with positive skill at the 95% confidence level over a 41-year period from early May. Owens and Landsea (2003) examined the skill of Gray’s operational Atlantic seasonal tropical cyclone forecasts relative to climatology and persistence. Their analysis indicated that for the analyzed period (1984 - 2001) both the statistical and the adjusted forecasts demonstrated skill over climatology and persistence. There is also evidence that the adjusted forecast was more skillful than the statistical model forecast. The TSR forecast in hindcast and operational mode is also skillful (Lea and Saunders, 2006) using as skill measure the mean square skill score (percentage reduction is mean square error compared to a rolling prior 10 year climatology). 4.3.1b. Landfall Probability Forecasts Seasonal forecasts of landfall probabilities for the Atlantic have been issued by Colorado State University since August 1998. These probabilities are based upon a forecast of tropical cyclone activity and a measure of North Atlantic SSTs. In general, when an active season was predicted, the probability of landfall was increased. The CSU forecast team has recently also calculated landfall probabilities for 11 regions, 55 sub-regions and 205 coastal and near-coastal counties from Texas to Maine (Klotzbach, 2006). The Cuban Meteorological Institute also issues statistical landfall forecasts of tropical cyclones in Cuba, based on a discriminant function methodology. In a recent paper (Saunders and Lea 2005), TSR describes their new forecast model for issuing in early August skilful seasonal predictions of hurricane landfall activity for the coast of the United States. The new prediction model uses wind patterns to predict the U.S. ACE index (effectively the cumulative wind energy from all U.S. striking tropical storms during the main hurricane season). The July height-averaged winds in these regions are indicative of atmospheric circulation patterns that either favor or hinder evolving hurricanes in reaching U.S. shores. The model gives forecasts from 1 August. 97% of all intense hurricane strikes on the U.S. and 87% of all hurricane hits on the U.S. occur after this date. The model correctly anticipates whether US hurricane losses are above-median or below-median in 74% of the years between 1950 and 2003. It also performed very well in ‘real-time’ operation in 2004 and 2005. For these damaging hurricane seasons the model predicted U.S. landfalling hurricane activity in the upper quartile (2004) and upper decile (2005) of years historically. 4.3.1c Dynamical Tropical Cyclone Seasonal Forecasts The IRI (International Research Institute for Climate and Society) and ECMWF (European Centre for Medium-Range Weather Forecasts) issue seasonal forecasts of tropical storm frequency based on dynamical models. The ECMWF forecasts are based on coupled ocean-atmospheric models (Vitart and Stockdale 2001). In contrast, the IRI forecasts are obtained in a two-tier procedure, by first forecasting various possible scenarios for the sea surface temperatures (SST) using statistical or dynamical models and then forcing the atmospheric models with those predicted SSTs. Tropical cyclone-like vortices are then identified and tracked in the atmospheric model outputs (e.g. Vitart 1997; Camargo and Zebiak 2002). The IRI also issues ACE forecasts based on dynamical models for a few Northern Hemisphere regions. The skill of some dynamical models to predict the frequency of tropical storms over the Atlantic can be comparable to the skill of statistical models. Over the other ocean basins, dynamical models can also display some robust skill in predicting the frequency of tropical storms, but they usually perform poorly over the North and South Indian oceans (e.g. Camargo et al. 2005a). It is not clear if this due to model errors or to a lack of predictability. Combining different model forecasts (multi-model ensemble forecast) seems to produce overall better forecasts than individual ensemble forecasts (Vitart 2006). The skill of

495

various climate models for seasonal tropical cyclone activity in hindcast mode is discussed in Camargo et al. (2005a) and Vitart (2006). The seasonal prediction of the risk of tropical storm landfall still represents a challenge for dynamical models. The tropical storm tracks in seasonal forecasting systems are usually unrealistically too poleward due to the too coarse horizontal resolution of the models. Either finer resolution or the use of statistical techniques such as clustering (Camargo et al. 2005b) would be needed to predict the risk of tropical storm landfall. 4.3.2 Sub-seasonal Tropical Cyclone Forecasts Interest in the prediction of atmospheric variability on the intra-seasonal timescale has recently blossomed (e.g., Schubert et al 2002, Waliser et al. 2006). On this timescale, the Madden-Julian oscillation (MJO), with its 30- to 80-day period, provides the greatest prospects for tropical prediction. Concurrent with the developments in MJO prediction, the modulation of TC activity by the MJO has been shown for many of the world’s major TC formation regions (e.g., Liebmann et al. 1994, Maloney and Hartmann 2000, Molinari and Vollaro 2000, Hall et al. 2001, Bessafi and Wheeler 2006). Thus there exists hope for practical application of the MJO for TC activity forecasting in the near future. MJO prediction has so far been approached using mainly empirical methods (see review by Waliser (2005)), owing to the difficulty that global numerical models have in its simulation and prediction (e.g., Jones et al. 2000, Lin et al. 2006). Useful predictive skill from empirical methods has been quoted in the range of 15 to 20 days for large-scale fields in the tropics. The crux of the empirical problem is the extraction of the MJO’s frequency-limited signal from observational data in real-time. Empirical methods then evolve this signal in a way that is consistent with the statistics of past MJO events. The first empirical method to be implemented in real time involves Fourier wavenumber-frequency filtering of daily updated outgoing longwave radiation (OLR) data (Wheeler and Weickmann 2001), available online since 2000. Filtered fields constructed for times after the end of the dataset are used as a skilful forecast, as applied to the MJO and other tropical waves as well. Despite being a forecast of large-scale OLR only, and not of TC activity, it has gained a broad awareness amongst tropical forecasters. The use of empirical orthogonal functions (EOFs) to extract the MJO’s signal has also gained common usage. The NOAA Climate Prediction Center produces ten indices for monitoring the different longitudinal stages of the MJO. The Australian Bureau of Meteorology computes the daily projection onto the leading pair of EOFs of the combined fields of equatorially-averaged OLR, 850-hPa zonal wind, and 200-hPa zonal wind, producing a two-component index of the MJO (Wheeler and Hendon 2004). WWW-sites provide the daily-updated indices as well as their historical values (back to the 1970s), and have been applied to the study of TC activity modulation (e.g., Wheeler and McBride 2005, Harr 2006) and prediction (Leroy et al. 2004). The modulation of TC numbers by the phase of the MJO has been quoted to be as high as 4:1 in some locations (e.g., Hall et al. 2001; Maloney and Hartmann 2000). TC genesis tends to preferentially occur near, and a little westward, of the longitude of maximum MJO convective activity. These are regions of enhanced low-level cyclonic vorticity associated with the near-equatorial convective forcing. Application of such information by TC forecasters has so far been mostly subjective. One objective method, however, is that developed by Leroy et al. (2004), which provides predictions of weekly probabilities of TC activity within large zones. Predictors are the MJO indices of Wheeler and Hendon (2004), large-scale patterns of SST as indicators of ENSO and interannual Indian Ocean variability, and the climatological seasonal cycle of TC activity. Verification statistics show that incorporation of the MJO predictor increases skill out to 2-3 weeks. Greatest skill is achieved during times when the MJO is strong, and little improvement is gained when the MJO is weak. Further increases in skill would be

496

expected through the incorporation of information from equatorial Rossby (ER) waves, and potentially other convectively-coupled equatorial waves as well (Leroy et al. 2004, Bessafi and Wheeler 2005, Frank and Roundy 2006), but given their lower explained variance and higher frequency, the predictability and lead-time they provide is less (Wheeler and Weickmann 2001). An empirical method that has sought to include the effects of a wide variety of wave modes and climate signals to forecast local daily probabilities of TCs has recently been developed by Paul Roundy at CIRES. It is based on research showing the relationships of the waves to TCs (Frank and Roundy 2006), and the wave’s self-consistent patterns of propagation and interaction (Roundy and Frank 2004a, 2004b). The method, employing logistic regression, fits a hyperbolic tangent function to the relationship between these modes and time series representing the local presence of TCs. The optimum subset of available predictors is found for each region and time lag. The method shows improvement of roughly 10-40 percent over climatological probabilities at one-week lead times, depending on location. Regions where skill tends to be highest include the Northeast Pacific, the Northwest Pacific, and the Bay of Bengal. This method acts like an analogue forecast because probabilities of a TC occurring in a given wave state depend on how often they formed in similar states in the past. Consequently, if a TC forms within a unique wave state or a wave state that has not often been associated with TCs in the past, high probabilities might not be forecast. The method is skillful because many TCs form within similar large-scale wave and climate states. While there is much room for improvement in the skill and application of empirical/statistical methods of intra-seasonal TC prediction, the greatest hope for improvement lies with dynamical/numerical models. Indeed, numerical studies using twin-experiment methodology in which the model employed is assumed to be perfect (Waliser et al. 2003), indicate useful predictability of the MJO may extend to 25-30 days, 10 days longer than that currently derived from empirical methods. Empirical methods are limited in the totality of the weather/climate system they can predict, their ability to adapt to arbitrary conditions, and their ability to take advantage of known physical constraints (Waliser 2005). Given the inadequacy of the representation of the MJO, and other tropical waves, in current numerical prediction models, however, much work remains to achieve their theoretical potential. A recent development in statistical tropical cyclone prediction in the Atlantic is the prediction of an individual month’s tropical cyclone activity by the Colorado State University team. These shorter-term (than seasonal) predictions are issued due to the fact that inactive seasons can have active months and active seasons can have inactive months. Individual monthly prediction began with a prediction of August-only activity issued with the 1 August seasonal forecast in 2000. Following the success of the August-only forecast (Blake and Gray, 2004), September-only (Klotzbach and Gray 2003) and October-only forecasts were developed. 4.3.3 Conclusions Statistical seasonal tropical cyclone forecasting has come a long way since it began in the early 1980s. Along with predictions of total seasonal activity, several forecasts now include individual monthly forecast and predictions of probability of landfall. As the availability of global datasets such as the NCEP/NCAR and ECMWF Reanalysis continue to be improved, so will statistical forecasts of tropical cyclone. An updated and homogenous quality best-track dataset globally would also contribute for more skilful forecasts. Dynamical seasonal tropical cyclone forecasts are now currently issued for various regions. Increasing model resolution should help improve the skill of these forecasts. In order to be able to forecast landfall probabilities using dynamical models, systematic biases in the tracks of model tropical cyclones need to be examined and explained. Some of the biases are probably not only due to low-resolution, and more research is needed in understanding the atmospheric models’ ability to forecasts tropical cyclones.

497

On sub-seasonal time scales, with the improvement of the dynamical and statistical models in forecasting MJO and other large-scale waves, forecasting tropical cyclone activity should be possible operationally at short lead times in the near future, as already occurs experimentally in a few cases. References Ballester, M., C. González, and R. Peréz Suaréz, 2004a: Modelo estadísitico para el pronóstico de la actividad ciclónica en el Oceáno Atlántico, el Golfo de México y el Mar de Caribe, Revista Cubana de Meteorología, Vol,. 11, No1, 9pp, in Spanish, available from biblio@met.in.cu . Ballester, M., C. González, R. Peréz Suaréz, A. Ortega, and M. Sarmiento, 2004b: Pronóstico de la actividad ciclónica en la region del Atlántico Norte, con énfasis en el Caribe y Cuba, Informe Científico, Instituto de Meteorología, in Spanish, available from biblio@met.in.cu . Bell, G. D., and M. Chelliah, 2006: Leading tropical modes associated with interannual and multi-decadal fluctuations in North Atlantic hurricane activity. J. Climate 19, 590-612. Bessafi, M., and M.C. Wheeler, 2006: Modulation of south Indian Ocean tropical cyclones by the Madden-Julian oscillation and convectively coupled equatorial waves. Mon. Wea. Rev., 134, 638-656. Camargo, S. J. and S. E. Zebiak, 2002: Improving the detection and tracking of tropical storms in atmospheric general circulation models. Wea. Forecasting, 17, 1152-1162. Camargo, S. J., A. G. Barnston and S. E. Zebiak, 2005a: A statistical assessment of tropical cyclones in atmospheric general circulation models. Tellus 57A, 589-604. Camargo, S. J., A. W. Robertson, S. J. Gaffney, P. Smyth and M. Ghil, 2005b. Cluster analysis of western North Pacific tropical cyclone tracks, IRI Technical Report 05-03, 57 pp., International Research Institute for Climate and Society, Palisades, NY, submitted to Journal of Climate (December, 2005). Chan, J. C. L., J. E. Shi, and C. M. Lam, 1998: Seasonal forecasting of tropical cyclone activity over the western North Pacific and the South China Sea. Wea. Forecasting, 13, 997-1004. Chan, J. C. L., J. E. Shi, and C. M. Lam, 2001: Improvements in the seasonal forecasting of tropical cyclone activity over the western North Pacific, Wea. Forecasting, 16, 997-1004. Frank, W.M., and P.E. Roundy, 2006: The role of tropical waves in tropical cyclogenesis. Mon. Wea. Rev., in press Goldenberg, S.B., C.W. Landsea, A.M. Mestas-Nuñez, and W.M. Gray, 2001: The recent increase in Atlantic hurricane activity: Causes and implications. Science, 293, 474-479. Gray, W. M., 1984a: Atlantic seasonal hurricane frequency. Part I: El Niño and 30 mb quasi-biennial oscillation influences. Mon. Wea. Rev., 112, 1649–1668. Gray, W. M, 1984b: Atlantic seasonal hurricane frequency. Part II: Forecasting its variability. Mon. Wea. Rev., 112, 1669–1683. Gray, W. M., C. W. Landsea, P. W. Mielke Jr., and K. J. Berry, 1992: Predicting Atlantic seasonal hurricane activity 6–11 months in advance. Wea. Forecasting, 7, 440–455.

498

Hall, J.D., A.J. Matthews, D.J. Karoly, 2001: The modulation of tropical cyclone activity in the Australian region by the Madden-Julian oscillation. Mon. Wea. Rev., 129, 2970-2982. Harr, P.A., 2006: Temporal clustering of tropical cyclone occurrence on intraseasonal time scales. 27th Conference on Hurricanes and Tropical Meteorology, Monterey, CA, 24-28 April 2006, Extended abstract 3D.2. Jones, C., D. E. Waliser, J.-K. E. Schemm, and K.-M. Lau, 2000: Prediction skill of the Madden Julian oscillation in dynamical extended range forecasts. Climate Dyn., 16, 273-289. Klotzbach, P. J., and W. M. Gray, 2003: Forecasting September Atlantic basin tropical cyclone activity. Wea. Forecasting, 18, 1109-1128. Klotzbach, P. J., and W. M. Gray, 2004: Updated 6-11 month prediction of Atlantic basin seasonal hurricane activity. Wea. Forecasting, 19, 917-934. Klotzbach, P. J., 2006: United States landfalling hurricane probability webpage, 27th Conference on Hurricanes and Tropical Meteorology, Monterey, 24-28 April, 10A.1. Lea, A. S. and Saunders, M. A., 2006: Seasonal prediction of typhoon activity in the Northwest Pacific basin, 27th Conference on Hurricanes and Tropical Meteorology, Monterey, 24-28 April, P. 5.23. Leroy A., M.C. Wheeler, and B. Timbal, 2004: Statistical prediction of the weekly tropical cyclone activity in the Southern Hemisphere. Internal report for the Bureau of Meteorology and Meteo France, 66pp. Liebmann, B., H.H. Hendon, and J.D. Glick, 1994: The relationship between tropical cyclones of the western Pacific and Indian Ocean and the Madden-Julian oscillation. J. Meteor. Soc. Japan, 72, 401-412. Lin, J.-L., G.N. Kiladis, B.E. Mapes, K.M. Weickmann, K.R. Sperber, M.C. Wheeler, S.D. Schubert, A. Del Genio, L.J. Donner, S. Emori, J.-F. Gueremy, F. Hourdin, P.J. Rasch, E. Roeckner, and J.F. Scinocca, 2006: Tropical intraseasonal variability in 14 IPCC AR4 climate models. Part I: Convective signals. J. Climate, 19, 2655-2690. Liu, K. S. and J. C. L. Chan, 2003: Climatological characteristics and seasonal forecasting of tropical cyclones making landfall along the South China coast. Mon. Wea. Rev., 131, 1650–1662. Lloyd-Hughes, B., M. A. Saunders and P. Rockett, 2004: A consolidated CLIPER model for improved August-September ENSO prediction skill, Wea. Forecasting, 19, 1089-1105. Maloney, E.D., and D.L. Hartmann, 2000: Modulation of hurricane activity in the Gulf of Mexico by the Madden-Julian oscillation. Science, 287, 2002-2004. Molinari, J. and D. Vollaro, 2000: Planetary- and synoptic-scale influences on eastern Pacific tropical cyclogenesis. Mon. Wea. Rev., 128, 3296-3307. Owens, B.F. and C.W. Landsea, 2003: Assessing the skill of operational Atlantic seasonal tropical cyclone forecasts, Wea. Forecasting, 18, 45-54. Roundy P. E., and W. M. Frank, 2004a: A climatology of waves in the equatorial region. J. Atmos. Sci. 61, 2105-2132. Roundy P. E., and W. M. Frank, 2004b: Effects of low-frequency wave interactions on intraseasonal

499

oscillations. J. Atmos. Sci. 61, 3025-3040. Saunders, M. A. and A. S. Lea, 2005: Seasonal prediction of hurricane activity reaching the coast of the United States. Nature, 434, 1005-1008. Schubert, S., R. Dole, H. van den Dool, M. Saurez, and D. Waliser, 2002: Prospects for improved forecasts of weather and short-term climate variability on subseasonal (2 weeks to 2 months) time scales. NASA Tech. Rep. NASA/TM 2002-104606, Vol. 23, 171 pp. Vitart F., J. L. Anderson and W. F. Stern, 1997: Simulation of interannual variability of tropical storm frequency in an ensemble of GCM integrations. J. Climate, 10, 745–760. Vitart F. D. and T. N. Stockdale, 2001: Seasonal forecasting of tropical storms using coupled GCM integrations. Mon. Wea. Rev., 129, 2521–2537. Vitart F., 2006: Seasonal forecasting of tropical storm frequency using a multi-model ensemble. Q. J. R. Meteorol. Soc. 132, 647-666. Waliser, D.E., 2005: Predictability and forecasting. In: W.K.M. Lau and D.E. Waliser (eds), Intraseasonal Variability in the Atmosphere-Ocean Climate System. Praxis, Springer Berlin Heidelberg, pages 389-423. Waliser, D. E., K.-M. Lau, W. Stern, and C. Jones, 2003: Potential predictability of the Madden-Julian oscillation. Bull. Amer. Meteor. Soc., 84, 33-50. Waliser, D., K. Weickmann, R. Dole, S. Schubert, O. Alves, C. Jones, M. Newman, H.-L. Pan, A. Roubicek, S. Saha, C. Smith, H. van den Dool, F. Vitart, M. Wheeler, and J. Whitaker, 2006: The experimental MJO prediction project. Bull. Amer. Meteor. Soc., 87, 425-431. Wheeler, M.C., and H.H. Hendon, 2004: An all-season real-time multivariate MJO Index: Development of an index for monitoring and prediction. Mon. Wea. Rev., 132, 1917-1932. Wheeler, M.C., and J.L. McBride, 2005: Australian-Indonesian monsoon. In: W.K.M. Lau and D.E. Waliser (eds), Intraseasonal Variability in the Atmosphere-Ocean Climate System. Praxis, Springer Berlin Heidelberg, pages 125-173. Wheeler, M., and K.M. Weickmann, 2001: Real-time monitoring and prediction of modes of coherent synoptic to intraseasonal tropical variability. Mon. Wea. Rev., 129, 2677-2694.

500

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES Topic 4a : Updated Statement on the Possible Effects of Climate Change on Tropical Cyclone Activity/Intensity Rapporteur: John McBride Bureau of Meteorology Research Centre Australia E-mail: J.McBride@bom.gov.au [Note from Co-chair, Johnny Chan: At the time of printing, we still have not received a written report from Dr. McBride on the points to be discussed that could lead to the development of an updated statement on the possible effects of climate change on tropical cyclone activity/intensity. Correspondence with Dr. McBride and Chair of Topic 4, Dr. Chris Landsea, suggested because Topic 4.2 contains a very detailed discussion on the state of science knowledge on possible relationships between climate change and tropical cyclone activity/intensity, it is not necessary for such a discussion to be repeated here. Instead, a report submitted to CAS in February 2006 by Dr. McBride as Chairman of TMRP's Steering Committee for Project TC-2: Scientific Assessment of Climate Change Effects on Tropical Cyclones is reproduced here (with very minor editing), which include a proposed statement on tropical cyclones and climate change, and could therefore form a basis for discussion at the Workshop.] Report from Chairman of TMRP's Steering Committee for Project TC-2: Scientific Assessment of Climate Change Effects on Tropical Cyclone Purpose: To provide an updated assessment of the current state of knowledge of the impact of anthropogenically-induced climate change on tropical cyclones Background: The WMO CAS Tropical Meteorology Research Program has undertaken a series of assessments of the potential influence of climate change on global tropical cyclone activity. The most recent was published in the Bulletin of the American Meteorological Society by Henderson- Sellers et al (1998) and had the following major conclusions:

• Whilst there was evidence of substantial multi-decadal variability (particularly for intense Atlantic hurricanes), there was no clear evidence of long-term trends;

• The Maximum Potential Intensity of cyclones will remain the same or undergo a modest increase of up to 10-20%. These predicted changes are small compared with the observed natural variations and fall within the uncertainty range in current studies;

• Little can be said about the potential changes of the distribution of intensities as opposed to maximum achievable intensity;

• Current knowledge and available techniques are not able to provide robust quantitative indications of potential changes in tropical cyclone frequency;

501

• The broad geographic regions of cyclogenesis and therefore also the regions affected by tropical cyclones are not expected to change significantly;

• The modest available evidence points to an expectation of little or no change in global frequency. Regional and local frequencies could change substantially in either direction, because of the dependence of cyclone genesis and track on other phenomena (e.g. ENSO) that are not yet predictable;

• The rapid increase of economic damage and disruption by tropical cyclones has been caused, to a large extent, by increasing coastal populations, by increasing insured values in coastal areas and, perhaps, a rising sensitivity of modern societies to disruptions of infrastructure.

A number of high-impact tropical cyclone events have occurred throughout the globe during 2004 and 2005, including:

• Ten fully developed tropical cyclones made landfall in Japan in 2004, causing widespread damage.

• Southern China experienced much below-normal tropical cyclone landfalls and subsequently suffered a severe drought;

• Four major hurricanes caused extensive damage and disruption to Florida communities in 2004;

• In March 2004 southern Brazil suffered severe damage from a system that had hurricane characteristics, the first recorded cyclone of its type in the region;

• Five fully-developed cyclones passed through the Cook Islands in a five week period in February-March 2005;

• The 2005 North Atlantic Hurricane Season broke several records including number of tropical cyclones, number of major hurricanes making landfall and number of category 5 hurricanes. In particular, the landfall of Hurricane Katrina at New Orleans and Mississippi caused unprecedented damage and more than 1300 deaths.

The IPCC Third Assessment Report concluded that "most of the observed warming over the last 50 years is likely to have been due to the increase in greenhouse gas concentrations." There is now additional supporting evidence for this conclusion. There is strong and growing evidence that a warming signal has penetrated into the global oceans over the past 40 years and was likely caused primarily by anthropogenic forcing. At the regional scale, sea surface temperatures in the major tropical ocean basins have warmed, with a likely substantial contribution from anthropogenic forcing indicated in several of the basins. Further, two scientific papers appeared during 2005 in highly visible journals (Nature and Science) providing evidence for an increase in the number of the intense cyclones. This combination of events has led to statements in the world press that the recent hurricane disasters can be directly attributed to the impact of global warming. Hence it is appropriate that the Tropical Meteorology Research Program Panel of Experts make a statement to guide member countries. The membership of this panel includes two of the authors of the Nature and Science papers, as well as prominent tropical cyclone researchers from USA, Australia, UK and China:

Dr. John McBride, Dr Jeff Kepert (Australia) Prof. Johnny Chan (Hong Kong, China) Julian Heming (United Kingdom) Dr. Greg Holland, Professor Kerry Emanuel, Dr. Thomas Knutson, Dr Hugh Willoughby, Dr. Chris Landsea (USA)

502

Statement on tropical cyclones and climate change: We consider that the following conclusions of Henderson-Sellers et al (1998) remain valid:

• Current knowledge and available techniques are not able to provide robust quantitative indications of potential changes in tropical cyclone frequency;

• The modest available evidence points to an expectation of little or no change in global frequency. Regional and local frequencies could change substantially in either direction, because of the dependence of cyclone genesis and track on other phenomena (e.g. ENSO) that are not yet predictable;

• The rapid increase of economic damage and disruption by tropical cyclones has been caused, to a large extent, by increasing coastal populations, by increasing insured values in coastal areas and, perhaps, a rising sensitivity of modern societies to disruptions of infrastructure.

However, further elaboration is required on the conclusions by Henderson-Sellers et al. relating to changes in cyclone intensity. Whilst there is substantial debate on this topic, we consider that the following statements can be made:

• No single high impact tropical cyclone event of 2004 and 2005 can be directly attributed to global warming, though there may be an impact on the group as a whole;

• Emanuel (2005) has produced evidence for a substantial increase in the power of tropical cyclones (denoted by the integral of the cube of the maximum winds over time) during the last 50 years. This result is supported by the findings of Webster et al (2005) that there has been a substantial global increase (nearly 100%) in the proportion of the most severe tropical cyclones (category 4 and 5 on the Saffir-Simpson scale), from the period from 1970 to 1995, which has been accompanied by a similar decrease in weaker systems.

• The research community is deeply divided over whether the results of these studies are due, at least in part, to problems in the tropical cyclone data base. Precisely, the historical record of tropical cyclone tracks and intensities is a byproduct of real-time operations. Thus its accuracy and completeness changes continuously through the record as a result of the continuous changes and improvements in data density and quality, changes in satellite remote sensing retrieval and dissemination, and changes in training. In particular a step-function change in methodologies for determination of satellite intensity occurred with the introduction of geosynchronous satellites in the mid to late 1970’s.

• The division in the community on the Webster et al and on the Emanuel papers is not as to whether Global Warming can cause a trend in tropical cyclone intensities. Rather it is on whether such a signal can be detected in the historical data base. Also it can be difficult to isolate the forced response of the climate system in the presence of substantial decadal and multi-decadal natural variability, such as the Atlantic Multi-decadal Oscillation.

• Whilst the existence of a large multi-decadal oscillation in Atlantic tropical cyclones is still generally accepted, some scientists believe that a trend towards more intense cyclones is emerging. This is a hotly debated area for which we can provide no definitive conclusion. It is agreed that there is no evidence for a decreasing trend in cyclone intensities.

Besides the above comments on tropical cyclone intensity, based on the published, literature it is appropriate to make the following additional statements:

• The geographical extents of the existing regions of cyclogenesis and the existing regions affected by tropical cyclones still are not expected to change significantly. However, superimposed on the multi-decadal fluctuations, interannual variations such

503

as the El Niño-Southern Oscillation (ENSO) are a major influence on cyclone development and the subsequent paths in most parts of the world. There is no consensus among current climate models regarding how ENSO variability may change in the future, although any such changes in ENSO would be expected to alter Tropical Cyclones regionally.

• In the context of changing regions of cyclogenesis, we note the debate concerning the hurricane-like system in the South Atlantic, but consider no conclusions can be made based on a single system.

• A robust result in model simulations of tropical cyclones in a warmer climate is that there will be an increase in precipitation associated with these systems (for example, Knutson and Tuleya, 2004). The mechanism is simply that as the water vapor content of the tropical atmosphere increases, the moisture convergence for a given amount of dynamical convergence is enhanced. This should increase rainfall rates in systems (viz tropical cyclones) where moisture convergence is an important component of the water vapor budget. To date no observational evidence has been found to support this conclusion; so no quantitative estimate can be given for the anticipated rainfall increase without further research.

• While demographic trends are the dominant cause of increasing damage by tropical cyclones, any significant trends in storm activity would compound such trends in damage.

• Projected rises in global sea level are a cause for concern in the context of society’s vulnerability to tropical cyclones. In particular for the major cyclone disasters in history the primary cause of death has been salt-water flooding associated with storm surge.

• Because of the problems of the tropical cyclone databases utilized for studies on trends in these extreme events, there is an immediate need to conduct an in-depth storm-by-storm reanalysis of tropical cyclones in all basins. Currently, a reanalysis is underway only for the Atlantic basin.

• The research issues discussed here are in a fluid state and are the subject of much current investigation. Given time the problem of causes and attribution of the events of 2004-2005 will be discussed and argued in the refereed scientific literature. Prior to this happening it is not possible to make any authoritative comment.

Further action: The Project TC-2 Committee will produce an update to the Henderson-Sellers et al paper documenting the state of the science, explaining the basic principles, and outlining the sources of disagreement. This will be presented for discussion and ratification at the Sixth International Workshop on Tropical Cyclones (IWTC-VI) to be held in Costa Rica in November 2006. Subsequently it will appear as a WMO TMRP Report and be submitted to the refereed literature.

504

References: Emanuel, K., 2005: Increasing destructiveness of tropical cyclones over the past 30 years. Nature, 436, 686-688. Henderson-Sellers, A., Zhang, H., Berz, G., Emanuel, K., Gray, W., Landsea, C., Holland, G., Lighthill, J., Shieh, S.-L., Webster, P., McGuffie, K. 1998: Tropical Cyclones and Global Climate Change: A Post-IPCC Assessment. Bull. Amer. Meteor. Soc., 79, 19–38. Knutson, T.R. and R.E. Tuleya, 2004: Impact of CO2-Induced Warming on Simulated Hurricane Intensity and Precipitation: Sensitivity to the Choice of Climate Model and Convective Parameterization. J. Climat., 17, 3477–3495. Webster, P.J., G.J. Holland, J.A. Curry, and H-R. Chang, 2005: Changes in tropical cyclone number, duration and intensity in a warming environment. Science, 309, 1844-1846.

505

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES Topic 5 : Disaster Mitigation, Warning Systems and Societal Impact Topic Chair: M.C. Wong Hong Kong Observatory 134A Nathan Road Kowloon Hong Kong E-mail: mcwong@hko.gov.hk Incorporating: Topic 5.1 Evaluating the Effectiveness of Warning Systems Rapporteur: LEE Woo-Jin (Republic of Korea) Topic 5.2 Factors Contributing to Human and Economic Losses Rapporteur: Roger A. PIELKE Jr. (USA) Topic 5.3 Mitigation Strategies and Community Capacity Building for Disaster Reduction Rapporteur: Linda ANDERSON-BERRY (Australia) 5.0 Introduction Topic 5 focuses on the application aspects of tropical cyclone forecasting and warnings, and the way such information is conveyed to stakeholders, users and the general public for the mitigation of adverse cyclone impacts. Key findings of the studies carried out by the working groups on the three sub-topics are summarized and progress since IWTC-V is highlighted, along with potential subject areas for discussion in IWTC-VI and recommendations to be made for future development initiatives. 5.0.1 Evaluating the Effectiveness of Warning Systems An effective warning system consists of two components: reliable forecasting of tropical cyclones and efficient conveyance of warning information. Forecasting of tropical cyclones Satellite data, particularly from microwave channels, and EPS (Ensemble Prediction System) guidance or consensus track forecasts are becoming more extensively used by the operational centres with, in some cases, support of nowcasting tools for landfalling tropical cyclones. Track forecasting based on both statistical and dynamical models has improved steadily for the range up to 72 hours, and position forecasts for tropical depressions are in general less reliable those for more intense cyclones. Statistical and dynamical guidance still need to be improved further, particularly for weak or compact cyclones, for cyclones with unusual track behaviour, mutual interaction in a multi-cyclone situation, and cyclones undergoing extratropical transition. Intensity forecast remains a challenge, for which NWP (Numerical Weather Prediction) guidance has a limited value and needs to be supplemented by conceptual models and/or statistical models. Rapid

506

changes in the translation speed, structure, and distribution of precipitation and wind during extratropical transition can substantially reduce the skill of medium range forecast of sensible weather downstream of a tropical cyclone. EPS is increasingly used in the forecasting of tropical cyclone track and intensity, particularly for the pre-alert of severe weather in the medium-range. EPS products are now increasingly available, and forecasters need to acquire skill and knowledge on how to interpret the EPS products, to assess the uncertainty associated with the EPS guidance, and to utilize the information for optimal operational benefits. Research to make EPS applicable to the forecasting of TC and related weather and hazards (landslide, storm surge, wave and flood forecasting) is being undertaken, and the outcome and experience to be shared with operational forecasters. Methods for calibration of TC intensity, utilization of consensus track forecast or an ensemble of EPS, interpretation techniques of probabilistic forecasts for decision makers, and downscaling techniques need to be developed. It is recommended that DPP (Disaster Prevention and Preparedness) effort should gear towards the prediction of extreme weather events. Studies on the potential impact of climate variability on the recurrence of extreme weather events should be carried out to enhance disaster preparedness. There still remains a technological gap among meteorological and hydrological centres, and international cooperation needs to be strengthened further to share information, resources and analysis tools. While various tools are available for use, opportunities for direct communication among scientists working in different disciplines such as meteorology, hydrology, oceanography, and other geosciences are still relatively limited. Large uncertainty exists in the forecasting of flooding, storm surge, and landslides as a tropical cyclone makes landfall. Conveyance of information In the conveyance of information, the uncertain nature of forecasts and potential vulnerability are increasingly reflected in the probabilistic expression of forecasts and warnings. Multimedia channels, including Internet, mobile phones, and digital multimedia broadcasting, are more widely used for dissemination. The intensity scales and advices on protective measures against tropical cyclone impact have been refined further to personalize the risk. Various outreach programmes have been effectively conducted to promote public awareness. Efforts are stepped up to enhance public awareness in cyclone forecasting through extensive funding investment and dedicated efforts of meteorologists and scientists on a worldwide scale. To improve public understanding of probabilistic prediction, one possible solution is to translate it into a more visual format for different stakeholders. Assistance from social scientists may be necessary in the translation process. Research should be conducted to find out what people understand and do not understand, and the results evaluated through demonstration projects. The communication channels for the dissemination of warnings continue to expand to DMB (Digital Multimedia Broadcasting), Internet portal sites, mobile phones, satellite broadcasting. Conventional media such as radio and TV remain an extremely effective and widely adopted means of disseminating weather information for both developed and developing countries, while door-to-door notification in local communities is considered an effective way of communication to personalize the risk. While NMHSs (National Meteorological and Hydrological Services) should take advantage of the advances in communication technology such as wireless broadband access, GPS(Global Positioning System), and GIS(Geographical Information System) to enhance the relevance and effectiveness of warnings, options and backup capabilities to disseminate warnings through multiple and diverse channels with a variety of high and low technology should be retained. Different forms of presentation of tropical cyclone messages should be developed, tailor-designed according to the strength of the dissemination channels and the level of the intended audience to enable effective communication of messages to different sectors of the community.

507

It is found that warnings conveyed and interpreted in terms of action words are more likely to trigger prompt responses from users. Warning messages should therefore contain specific information on what is to be done by individuals and by the community as a whole to minimize loss of life and property. To enhance response effectiveness through personal interaction, attempts should be made to engage volunteers at the community level under appropriate organizational arrangement as messengers for tropical cyclone warning dissemination on a door-to-door basis. Such initiatives undertaken in a cost-effective manner and with a high degree of community involvement can contribute positively to disaster mitigation efforts. Despite advances in tropical cyclone forecasting and communication technology, a problem persists with the public not responding adequately to the threat of tropical cyclones, even when the warning itself is accurate and disseminated in time. Numerous factors influence the behaviour of users, including previous experience of false alarms, rare occurrence, the belief that “it won’t happen here”, and so on. Various activities are ongoing to enhance public awareness on the risk of tropical cyclones through outreach programmes, primary and secondary school education, as well as cooperation among decision makers, emergency managers, media, and stakeholders down to the community level. NMHSs are encouraged to collaborate with social scientists, researchers and other stakeholders to develop TC disaster scenarios and visualize hazards in the form of hazard map, risk map or disaster management map, designating evacuation sites and displaying warning/evacuation signs etc. to maximize the effectiveness of public education initiatives. Social impact study of tropical cyclones should be conducted on a periodic basis by a joint team of meteorologists, hydrologists and social scientists in cyclone prone communities in line with the recommendation of IWTC-V. In societies with dense population, the pressure on land use plus the attraction of coastal land for agricultural or recreational purposes attracts people to settle in vulnerable areas or exposed islands along the coast. Regulatory framework on land use in the coastal region should be developed and enforced to reduce vulnerability. 5.0.2 Factors Contributing to Human and Economic Losses To evaluate losses through the years, an indexation (normalization) methodology incorporating factors to account for changes in population, inflation, and wealth is desirable. Indexed tropical cyclone losses are further adjusted, where appropriate, to account for the greatly improved building standards that may have been put into place in cyclone-prone areas. Given the significance of societal changes in trends of cyclone damage, another way to present a more accurate perspective on such trends is to consider how past events would affect present society. After the damage is normalized using changes in inflation, increases in population and economic activity in the region, no consistent trend is found in the normalized damage values according to cyclone impact studies in India, U.S., the Caribbean, and Latin America, implying that the key factors contributing to increased losses are largely attributable to societal factors. Increased attention to the need for rigorous cost-benefit analyses of disaster mitigation policy alternatives and practices is recommended. Irrespective of how such studies turn out (in terms of relative costs and benefits), there is a substantial benefit to decision making related to disasters to be gained from a more comprehensive and rigorous understanding of the value of disaster mitigation. More fundamentally, there is also a pressing need for more thorough information on the broad human and economic impacts of disasters, as well as indicators of relative vulnerability in order to help prioritize disaster mitigation investments. Re-insurance companies have begun discussions about the creation of an open-source database. The tropical cyclone community is recommended to collaborate on such efforts to create a centralized, comprehensive, peer-reviewed reference database. In

508

particular, significant by their absence, data on the non-economic human losses related to tropical cyclones are not always readily available. New and innovative policy options have been proposed for disaster mitigation that will likely stimulate demand for greater attention to costs and benefits. Among these are the securitization of risk through financial products such as catastrophe bonds and derivatives and the provision of micro-finance in developing countries as a tool for disaster recovery in ways that reduce long-term vulnerabilities. Such policies are not widespread, however, and they have not been subject to rigorous evaluation of costs and benefits, but nonetheless have strong support among many disaster experts. Besides the strong impact of increasing exposures and vulnerabilities on increasing losses due to societal changes over time, the possibility cannot be ruled out that an increase in SST (sea surface temperature) due to anthropogenic climate change would in the long term also lead to more tropical cyclone damage and losses. However, depending on the statistical approaches adopted, scientific evidences are so far inconclusive owing to the complex sequence of relationships linking SST, TC occurrence, TC behaviour, TC landfall impact and social-economic damage. 5.0.3 Mitigation Strategies and Community Capacity Building for Disaster Reduction Despite ever-improving technological solutions for forecasting tropical cyclones and communicating warning messages, human suffering and social, economic and environmental loss as a result of landfalling tropical cyclones continues to increase. Additionally, in the face of changing global climate regimes and the likelihood of more frequent, and possibly more intense, tropical cyclones, coupled with growing populations in tropical coastal regions, it is likely that more people in cyclone prone regions will be in harm’s way and the level of loss and suffering will continue to escalate. In view of such an undesirable trend, warnings systems have been developed throughout the WMO community in the context of a total warning system. More than just a matter of simple delivery of a message about impending severe weather conditions, this extends the concept of warnings to a complete end-to-end process that begins with using the best science available to predict and monitor the development and progress of cyclones all the way to the production and delivery of timely and accurate messages to a receptive, prepared and resourceful community in a format that is well understood. As such, identifying, understanding and reducing community vulnerabilities are just as important as disaster mitigation solutions gradually shift towards the development of social policies as well as engineering of defences. This direction was reflected in presentations at IWTC-V and a strategy has since evolved into the development and implementation of the Hyogo Framework for Action 2005-2015 – Building the Resilience of Nations and Committees to Disasters and the Platform for the Promotion of Early Warnings (PPEW). ‘Community capacity’ is increasingly being recognized as a reliable indicator of how human populations are likely to respond to and recover from the impact of disastrous events. Weather services, community and emergency managers, academics and hazard research centres are all increasingly investing in research and activities aimed at investigating community capacity, and ultimately applying strategies to build community capacity in support of disaster mitigation. The goal is to support a ‘bottom-up’ approach to strengthen and build resilient communities with the capacity to prepare for, mitigate, respond to and recover from natural disasters. Mitigation strategies in the context of community capacity building are illustrated using examples from Australia, Hong Kong, Fiji and the RANET (RAdio & InterNET for the Communication of Hydro-Meteorological and Climate Related Information) project. One of the outcomes of the IWTC-V was to cooperatively and collectively evaluate the effectiveness of tropical cyclone warning systems and the impact of landfalling tropical cyclones on coastal

509

communities. It was agreed that this should ideally be achieved through international and multi-disciplinary post-impact case studies. Results of case studies for Tropical Cyclones Zoe (Solomon Islands in December 2002) and Larry (north Queensland coast of Australia in March 2006) will be presented in IWTC-VI. 5.0.4 Conclusion Based on the above, it is recommended that the following key issues be further discussed in IWTC-VI with a view for future action: (a) the need to enhance knowledge and skills in TC intensity forecasting; (b) more research and development work on the applications of EPS, particularly for probabilistic

assessment of extreme events; (c) international effort to collate and compile a library of case studies on the effectiveness of TC

warning systems; (d) facilitation of more interaction and direct communication amongst meteorologists, hydrologists and

DPP experts; (e) promulgation of “ early warnings” as an “end-to-end” process, all the way through to the last mile to

get the messages to the “man-on-the-street” through enhanced community involvement; (f) adoption of the Hyogo Framework for Action (HFA) in the formulation of mitigation strategies; (g) establishment of a comprehensive reliable international “reference database” on the economic and

non-economic losses due to TC and related disasters; and (h) coordination of multi-disciplinary studies on the social impacts of TC and related disasters.

510

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES Topic 5.1 : Evaluating the Effectiveness of Warning Systems Rapporteur: Woo-Jin Lee National Meteorological Center, Korea Meteorological Administration, Seoul, Republic of Korea Email: wjlee@kma.go.kr or heseong@hanafos.com fax: (82-2) 2181-0629 Working Group: Peter J. Bowyer, Wenjie Dong, Charles Guard, Edwin S.T. Lai, W.-J. Lee, Nobutaka Mannoji, M Alimullah Miyan, Rosa Perez, T. Prasad, Alan Sharp, J. Weyman Abstract The warning process consists of two stages; forecasting and conveying information. Since IWTC-V, the guidance from ensemble prediction system and satellite observing system extensively used for the operational centers with the support of nowcasting tools for the landfalling tropical cyclones. The statistical and dynamical guidance yet need to be improved particularly for the weak cyclones, for the unusual track behaviors, and cyclones under extratropical transition. The uncertain nature of forecasts and increasing vulnerability has been reflected in the probabilistic expression of forecasts and warnings. The multimedia channels are more widely used for dissemination, including Internet, mobile phones, and digital multimedia broadcasting. The intensity scales and instruction to protect against a tropical cyclone have been refined further to personalize the risk. The various outreach programs has been effectively conducted to promote public awareness. Increased investment is required in public awareness in cyclone forecasting through very large investment of fund and dedicated efforts of meteorologists and scientists on a worldwide scale. There exist a technical gap among meteo-hydrological centers, and international cooperation need to be strengthened to share the information and analysis tools among the centers. 5.1.1 Introduction 5.1.1.1 The Team on Topic 5.1 on evaluating the Effectiveness of Warning System consists of 11 members. This draft is to summarize the major findings and recommendations on the topic based on the input from the members, and partially from the summary and recommendation at the workshop on effective tropical cyclone warning (Typhoon Research Coordination Group, 2005). 5.1.1.2 The effective measures for disaster preparedness is a well-functioning early warning system that delivers accurate and user-friendly information in a timely manner, considering the following aspects, as pointed out in the WMO Expert Meeting on Effective Early Warnings of Tropical Cyclones in Kobe, Japan on 17 and 18 January 2005, in association with the World Conference on Disaster Reduction (WCDR) (Kobe, 18 to 22 January 2005). The major elements are: (a) Adequate resources for disaster mitigation caused by tropical cyclones/severe weather hazards; (b) Improved accuracy in meteorological and hydrological forecasts for longer-ranges and quantification of uncertainty; (c) Qualified meteorological, hydrological and disaster prevention and preparedness personnel; (d) Sufficient attention to non-structural (public awareness, information sharing, etc.) mitigation

511

measures to cope with tropical cyclones/severe weather events; (e) Adequate institutional and infrastructure practices for coordination and capacity-building at national, regional and international levels to cope with the negative impact of tropical cyclones/severe weather risks on economic growth and human progress; (f) Adequacy of a National Disaster Management Policy that includes effective local dissemination of information to cope with the menace of meteorological and hydrological disasters; (g) Community consciousness for all stakeholders involved in tropical cyclone/severe weather-related disaster mitigation process and measures. 5.1.1.3 This draft is focused on the technical progress on one hand, and on the disaster prevention and preparedness on another, considering the viewpoints in the said WMO meeting. 5.1.2: Accuracy of track and intensity forecasts 5.1.2.1 The basic concerns for the forecasting of a tropical cyclone are wind, precipitation, storm surge and high wave, which depend on the track and intensity. Both dynamical and statistical models are used for the preparation of forecast at the pre-warning stage. 5.1.2.2 The new CLIPER extends the forecasts from 3 to 5 days and exhibits smaller forecast biases than the previous CLIPER, although forecast errors are comparable (Aberson and Sampson, 2003). An e-folding time-scale of about 15 h was calculated for the northwest Pacific basin, compared to just under 15 h in the Australia basin, and near 15 h in the North Atlantic. With e-folding error growth on that scale, 5-day forecasts can be expected to show some skill. As a result, interest has increased in medium-range tropical cyclone track prediction in the northwest Pacific basin, and the current operational at the Joint Typhoon Warning Center CLIPER model provides a baseline by which the track prediction skill can be measured. 5.1.2.3 The minimum attainable forecast error from an optimum statistical model with “perfect” input data are 53, 107, and 145 nm (98, 198, 269km) at 24, 48, and 72 hours respectively (OFC, 1997). 5.1.2.4 The performance of tropical cyclone track forecasts by leading centers has been continuously improving in recent years. (a) Operational track forecasts at RSMC Tokyo Typhoon Center for 19 tropical cyclones which attained TS intensity or higher in 2005 (as of 30 September) were verified against best track data of the Center. The annual mean position errors for this year are approximately 100 km (125 km in 2004) for 24-hour forecast, 174 km (243 km) for 48-hour forecast and 278 km (355 km) for 72-hour forecast. The annual mean position errors for 24-hour forecast in 2005 are smallest after each forecast started operationally. The annual mean ratios of EO (position errors of operational forecasts) to EP (position errors of PER-method forecasts) are 51 % (54 % in 2004) for 24-hour forecast, 37% (47 %) for 48-hour forecast and 37% (45%) for 72-hour forecast, which are also lowest after inauguration of each operational forecast. (RSMC Tokyo, 2005) (b) The recent assessment of tropical cyclone track forecasts showed that the 72-hr forecast error in the prediction of the tropical cyclone center in the South -West Indian Ocean is about 250 km. > 3.6 (RAI Tropical Cyclone Committee, 2005). (c) The global model at Korea meteorological Administration (KMA), the 72-hour model forecast error has decreased from roughly 500 km to roughly 400 km over the past 5 years. 5.1.2.5 Position forecast of a Tropical Depression is less reliable than that of a Tropical Storm. 5.1.2.6 Intensity is more difficult to forecast than position. Better understanding of the initiation and mechanisms of rapid deepening and of the timing and amount of intensity fluctuations caused by concentric eyewall cycles is essential to skillful intensity forecasts (OFC, 1997).

512

5.1.2.7 The potential forecast capabilities of STIPS model in terms of percent variance explained (R2) and mean absolute error (MAE) can be estimated from the dependent data. MAE is shown to increase from a value of 5.6 kt at 12 h to a value of 21.8 kt at 120 h. The percent variance explained of DELV, a measure of intensity change from the initial forecast time, starts with a relatively large value of 40 % at 12 h and increases to only 67.8 % at 120 h, keeping in mind that most (increasing 19.6%) of the variance of this variable is explained during the 12-60 h forecast time verses the 72-120 h forecast time (increasing 8.2 %). In independent predictions, these statistics are expected to degrade due to the influences of artificial statistical skill and the errors associated with the perfect prog assumption – particularly when track deviations are larger than 200 km (Knaff et al., 2005). 5.1.2.8 The annual mean RMSEs of central pressure forecasts at RSMC Tokyo Typhoon Center were 12.2 hPa (11.4 hPa in 2004), 16.4 hPa (16.1 hPa), and 19.4 hPa (18.6 hPa) for 24-, 48- and 72-hours, respectively, while those of maximum wind speed forecasts for 24 hours were 5.5 m/s (5.1 m/s in 2004), 7.5 m/s (7.1 m/s) and 10.4 m/s (8.1 m/s), respectively. (RSMC Tokyo, 2005) 5.1.2.9 Mesoscale numerical prediction for local wind and precipitation has not been improved significantly. Only a limited number of models have attained resolutions where cyclone structure (including intensity) can be addressed. These models are displaying real skill with regard to motion prediction and have the potential to handle cyclone genesis. (RAI Tropical Cyclone Committee, 2005). 5.1.2.10 The Tropical Cyclone Program had engaged the services of Systems Engineering Australia Pty. Ltd. in July 2003 to undertake reviews and assessments that would lead to suitable conversion factors between the WMO 10-minute average wind and 1-minute, 2-minute and 3-minute sustained winds. (RAV Tropical Cyclone Committee, 2004). 5.1.2.11 There is no commonly accepted definition of extratropical transition (ET). A variety of factors are assessed by different forecast centers to decide whether or not a tropical cyclone is undergoing ET. 5.1.2.12 Operational forecasting centers may continue to use the name assigned to the tropical cyclone during ET so that the general public does not underestimate the hazards associated with an ET event (e.g., in Canada an ET system is referred to as ‘‘post tropical cyclone’’). (Jones et al., 2003) Another extreme is to keep naming tropical cyclone while the ET cyclone is weakened sufficiently. 5.1.2.13 A presentation from Pat Harr and Sarah Jones at IWET2 in Halifax in 2003 indicated that the greatest model errors, globally, came from ET events. An ET event can substantially reduce the skill of the medium-range forecasts downstream of the tropical cyclone and, thus, can have an impact on Europe and western North America. Extratropical transition poses an especially challenging quantitative precipitation forecasting (QPF) problem. Successful QPF requires an accurate prediction of the track, intensity, and structural changes of storms undergoing ET. The timing of the precipitation shift relative to the storm track described above is very sensitive to physical mechanisms that govern the ET process (e.g., the dynamical and thermodynamic structure of the upstream trough). (Jones et al., 2003). 5.1.2.14 The rapid change in translation speed decreases the warning time for small fishing and recreational vessels that frequent the marine areas in summer and autumn. If the timing of the increase in translation speed is misjudged, track errors of hundreds of kilometers can occur. (Jones et al., 2003) 5.1.2.15 Use of the Cyclone Phase Space (CPS, Hart and Evans) diagrams as a diagnostic tool has proven invaluable in interpreting ET in various dynamical models (Hart, 2003). The Canadian Hurricane Centre has been successfully using this tool operationally since 2001. It is also an excellent tool for assessing tropical cyclone genesis within models.

513

Summary 5.1.2.16 The track forecasting based on both statistical and dynamical models have been improved steadily for the range up to 72 hours, but with further investigation on its performance in the range between 72-120 hours. The position forecast for a tropical depression is less reliable than that of a tropical storm. 5.1.2.17 The intensity forecast is a challenging subject in the operational center. The model guidance has a limited value, and supplemented by the conceptual models and or statistical models. 5.1.2.18 The rapid change in the translation speed, structure, and distribution of precipitation and wind during the ET can substantially reduce the skill of medium range forecast of sensible weather in the downstream of a tropical cyclone. From the public standpoint, the naming of ET may cause underestimation of the hazards ahead. Recommendation 5.1.2.19 Develop an integrated network for sharing of enhanced observations (GEOSS), model forecasts and products at the regional and global levels; 5.1.2.20 Use of more observational data improves short range forecast. Making the most of observational data and output of NWP model is necessary. Meteorological data such as raingauge data and radar data observed by organizations other than NMHSs should be obtained and utilized in order to monitor weather more effectively and to improve weather forecast. Raingauge data by volunteers could also be useful. Increased effort should be made to incorporate new and emerging data sources into dynamical model runs (data such as: dropsondes; satellite data such as QuikScat, AMSU, etc.) 5.1.2.21 Availability of these products based on the statistical and dynamical models, though, does not guarantee accuracy. The final warning may depend on the skills and experiences of duty forecasters. Hence familiarity of the forecasters with the performance of such tools will affect the decision making process of which product or combination of products will be the final basis of the warning 5.1.2.22 Continued interaction through workshops and advanced forecaster training sessions should be encouraged to develop national TC forecasting capability. 5.1.2 23 Greater awareness of the CPS diagrams should be generated and encouraged where dynamical models are used to assist forecasters. 5.1.2.24 Public understanding has to be enhanced to realize the hazards associated with Et cyclones as dangerous as s tropical cyclone 5.1.3 Interaction with other systems 5.1.3.1 Tropical cyclones often interact with other circulation systems such as monsoon, midlatitude troughs, and topography, etc.. Heavy rainfall and flooding is often associated with the remnants of tropical cyclones interacting with monsoon circulation in Philippines, Vietnam, and Laos. Recently, heavy rainfall of 300~500mm occurred on the stationary front intensifying with the input stream of moistures from the remnants of tropical cyclones, BILIS and KAEMI, at Korea. The two tropical cyclones made landfall at China before the heavy rainfall over Korea. 5.1.3.2 Large uncertainty in rainfall prediction is associated with track errors, interactions with other weather systems, and topography. The local dams or levee failure, debris in channels and streams, and

514

debris backup behind bridges may increase the possibility and scale of flash flooding. 5.1.3.3 Hydrologists need forecasts of the positions of tropical cyclones and the associated precipitation for various purposes. The required horizontal resolution, temporal resolution and lead time of the precipitation forecast depend on the specific purpose as well as the scale of the river. The precipitation forecast in the time scale of 1 day with the lead time of 3 days is required for the purpose of water resources management, drought relief and flood control for large basins. On the other hand, rainfall forecasts with intervals of 6 hours or less are required for flash floods, landslides, sediment disasters, and other hydrological requirements. 5.1.3.4 U.S. Geological Survey (USGS) and NOAA scientists use a variety of sophisticated operational hydrologic models for flood prediction, which could be adapted for use in data-poor settings, such as the island of Hispaniola. Hydrologic models will take the model- or satellite-estimated precipitation to create maps of flooding, a process now underway in the USGS Mekong River Project (Negri et al., 2005). 5.1.3.5 The current generation of landslide and debris-flow models is probabilistic because the scale of the phenomenon is small compared to the resolution of the currently available rainfall estimates. Nonetheless, simple solutions, such as the intensity-versus-duration plots, offer discrimination of landslide versus nonlandslide rainfall conditions for Puerto Rico (Negri et al., 2005). The soil water index developed at JMA, based on a tank model, is effective in operational use as an indicator for the prediction of landslides. Summary 5.1.3.6 Large uncertainty exists in the forecasting of flooding, storm surge, and landslides, as a tropical cyclone makes landfall. 5.1.3.7 Various tools are available, however, the communication need to be enhanced among scientists in the various disciplines including meteorology, hydrology, oceanography, and other geosciences. Recommendation 5.1.3.8 Direct communications between Meteorological Service and Hydrological Service should be engaged in real time for hydrologists to understand the probability of TC forecasts and for meteorologists to understand the requirements of hydrological issues. 5.1.3.9 Improve flash flood forecasting for ungauged basins and rainfall forecasts. 5.1.3.10 Improve coordination among meteorologists, hydrologists, DPPs, and other stakeholders for flash flood forecasting. 5.1.3.11 Considerable sophistication has been gained in storm surge forecasting. This gain should be reflected in providing more specific information regarding surge height and location warnings to permit effective evacuation. 5.1.3.12 Flash flood vulnerability indicators such as hazard mapping should be developed considering the urbanization and extension of megacities. 5.1.4 Application of ensemble prediction system (EPS) 5.1.4.1 TC forecasting and the associated decision making process involves high uncertainty. Ensemble Prediction Systems (EPS) provides extreme scenarios on severe weather phenomena, and

515

they serve as pre-alerts to forecasters well ahead of time, and inputs to other models such as storm surge and wave model, flood forecasting model, etc. 5.1.4.2 Krishnamurti et al. (1999) and Kumar et al. (2003) proposed that, by applying a simple multiple regression procedure that regresses different model forecasts against observations, a statistical prediction (a “superensemble”) can be obtained to give a better prediction than any of the model forecasts. Results from both tests indicated that the superensemble was able to produce an average 72-h TC track forecast error of ~210 km in the Atlantic (versus a consensus forecast error of ~300 km) and an intensity error of 20 kt (versus an ensemble average of five models of ~25 kt). 5.1.4.3 EPS has been operated in many NWP centers in recent years. Several meteorological centers provide their EPS products on the Internet. For instance, ECMWF open probability strike map on the web for WMO Members. Since the horizontal resolution of the global NWP models used in the EPS is relatively low, the models cannot always represent TC. Therefore, a calibration method to reduce underestimation of TC intensity should be developed. 5.1.4.4 The typhoon forecasts of nine Numerical Weather Prediction (NWP) centres are now posted in Web site at JMA and that they would shortly provide the ensemble mean of these forecasts. (RAV Tropical Cyclone Committee, 2004) 5.1.4.4 Ensemble techniques and probabilistic forecasts, either in the form of NWP EPS or multi-model ensembles, are probably most useful at the pre-warning stage. Under a cooperative research project between the Hong Kong Observatory and Japan Meteorological Agency (JMA) on the utilization and verification of JMA’s Ensemble Prediction System (EPS) tropical cyclone track data, a suite of new techniques including simple ensemble mean, track clustering, cluster means, conditioned strike probability and intensity calibration was developed. For divergent EPS track scenarios, conditioned strike probability using actual TC positions (i.e. positions observed between the last model run and the latest fix) a posteriori and adopting a clustering approach were found to have potential operation benefits. 5.1.4.5 The application extracts DMO (Direct Model Output) TC intensity information, categorizes the TC in terms of TD, TS, STS and T in a probabilistic sense and generates estimates of minimum MSLP and maximum winds at the centre of the TC. Using the TC category probabilities as weights, a probabilistic estimate of overall minimum MSLP and maximum can be derived. The technique as applied to JMA GSM has been completed. Extending the technique to other models will continue and eventually, combined estimates based on a multi-model approach are also feasible. 5.1.4.6 TC warning signals are issued based on the wind strength experienced or expected over Hong Kong. A methodology using a combination of statistical and NWP techniques has been developed to predict the probability of strong or gale force winds in Hong Kong. Occurrence probabilities are extracted based on climatological data from historical cases in 1968-2001 with respect to the TC’s position relative to Hong Kong and stratified by TC intensity (i.e. TD, TS, STS and T). The uncertainties in TC positions forecast by HKO’s operational warning track, which is very much guided by the multi-model ensemble forecasts, are estimated by: (i) past statistical position error distribution; or (ii) spread of individual model forecasts from the model ensemble mean compiled from ECMWF, JMA, NCEP and UKMO global models. A total of 25 member tracks are then constructed within 85% of the bivariate Gaussian distribution fitted to the perturbation fields. The probability of strong or gale force winds computed at hourly intervals is taken to be the weighted mean of probabilities for the 25 member tracks 5.1.4.7 The emergence of model ensemble techniques (for tropical cyclone prediction in the order of days) and nowcasting (for impact of inclement weather associated with tropical cyclones in the order of hours) are offering possibilities for forecasters to assess objectively the merits of various pre-warning and warning strategies. It is reasonable to assume that improved pre-warning and warning

516

procedures would be reflected in the long term assessment of benefits and gains as mentioned above. For tangible cost-benefit analysis, reduction in economic losses can then be evaluated against the cost of, say, maintaining 5 separate models or running the same model 50 times with different perturbations (to generate the model ensemble forecast), or the cost of operating a radar or local AWS network in support of various nowcasting systems. Summary 5.1.4.8 EPS is increasingly used in both track and intensity forecasting of tropical cyclone, particularly for the pre-alert of severe weather in the medium-range. 5.1.4.9 Many EPS products are available on the Web, and forecasters need to get information and knowledge on how to access, to interpret, and to estimate the uncertainty associated with the EPS guidance. Recommendations: 5.1.4.10 Outputs of EPS operated by NWP centers should be made available to NMHSs for TC forecasting. 5.1.4.11 Techniques to make EPSs applicable to TC forecasting should be developed. Methods for calibration of TC intensity, utilization of an ensemble of EPSs, interpretation techniques of probabilistic forecasts for decision makers, and downscaling techniques should be developed. 5.1.4.12 The potential benefits of EPS have to be realized through collaboration among meteorologists and various users. 5.1.4.13 The ongoing projects under Asian THORPEX and WMO demonstration projects have to be accelerated through regional cooperation 5.1.4.14 Once EPS is shown to be operationally useful, seminars and workshops for the training of forecasters to be acquainted with EPS should be held. Further recommendations are that at least two demonstration projects be held for the application of EPS in storm surge, landslide, wave model, and flood forecasting. 5.1.4.15 Hydrologists should explore the use of ensemble model outputs in flood forecasting. 5.1.5 Application of satellite observations and nowcasting tools 5.1.5.1 The fix of TC, and determination of the radius and deepness, based on the satellite observation and/or other in situ observation. is the very first step for the analysis of TC, and for the initialization of dynamical models. 5.1.5.2 Tropical cyclone fix has uncertainty especially in the early stages. In many cases of tropical cyclone formations over remote ocean areas, Quikscat gives the first evidence of the surface-wind circulation centre, and this sometimes results in major relocations of the centre. The Quikscat surface wind distributions have also improved the analysis of the outer vortex structure (e.g. 35 kt wind radius),which is useful for estimating ocean-surface wave generation, and wind-structure changes during extra-tropical transition of a tropical cyclone. A variety of opinions have been expressed as to the maximum wind speeds in a tropical cyclone that can be reliably measured by Quikscat. (Elsberry and Velden, 2003). 5.1.5.3 An Advanced ODT (AODT) and a multivariate linear regression technique that are being

517

developed by UW-CIMSS were presented in a special focus session at the IWTC-V. The AODT is being tested at the US National Hurricane Center and Joint Typhoon Warning Center and provides, specifically, intensity estimates for the tropical storm and tropical depression stages. If the AODT and regression technique do well in pre-operational testing, they would complement the manual Dvorak-type intensity estimates that nearly all warning centres use. Even small warning centres with PC-type computers could apply these two techniques. (Elsberry and Velden, 2003) Similar technique has been adopted at KMA as a test basis, which show positive impact during this season. 5.1.5.4 The microwave imagery is also useful for precipitation estimation, although only a few warning centres have taken advantage of this aspect. Whereas 21 of 31 centres indicate they had access, eight centres indicated that the microwave information was not used. Lack of training was cited as a primary reason why it was not used Questions also arise as to how to interpret the rain-flagged wind estimates and the various wind-direction ambiguity solutions. (Elsberry and Velden, 2003) 5.1.5.5 Tropical cyclone gale radii are routinely assessed by QuikScat, although its use for storm centre quantification is still problematic. 5.1.5.6 In the North Pacific basin, the definition of tropical storm is not consistent from center to center. For instance the RSMC-Tokyo and JTWC have different criteria for the discrimination of tropical storm from tropical depression, and for the definition of tropical depression. These may be of confusion for the definition of tropical depression itself. 5.1.5.7 A major advance in the past four years has been the availability on Internet Websites of real-time satellite imagery and digital data geo-referenced to the tropical cyclone position. Two of the best-known USA Websites are at the UW-CIMSS and the Naval Research Laboratory-Monterey (NRL). It has become evident that these Websites are being routinely accessed by many tropical cyclone warning centres. (Elsberry and Velden, 2003) 5.1.5.8 Nowcasting techniques as applied to the close approach and landfall of tropical cyclones are most relevant in shaping the warning strategies at the warning stage. The applications used at the Hong Kong Observatory for TC-related weather in terms of wind and rain is TC-LAPS and SWIRLS: 5.1.5.9 TC-LAPS is an application adapted for the nowcast of tropical cyclone winds from the Local Analysis and Prediction System (LAPS) of NOAA Forecast Systems Laboratory. Through a combination of successive correction and 3-D variational techniques, the system ingests conventional data as well as observations from less conventional sources such as AMDAR, QuikScat, Doppler winds, TREC winds (from SWIRLS, see below), profilers and AWS networks in an attempt to re-construct a 3-D wind fields for TCs approaching Hong Kong. The horizontal resolution is down to 1 km and the analyses are updated hourly. The nowcast is achieved by extrapolation along a given forecast track to generate time series of wind speed, wind direction and mean-sea-level pressure at specific locations in Hong Kong 5.1.5.10 SWIRLS (Short-range Warning of Intense Rainstorms in Localized Systems) is a radar-based QPF system designed for the nowcast of rainstorms in the next three hours and updated every six minutes. Radar-rain intensity is dynamically calibrated in real time based on ground truth from a dense network of raingauges. The extrapolation is achieved by the use of TREC (Tracking Radar Echoes by Correlation) winds. Results have shown that SWIRLS QPF is particularly effective from rain associated with TCs due to the dominant advective factor and relatively small fluctuations in rain intensity during the nowcast period. 5.1.5.11 By the time a TC hits Hong Kong, it is invariably near the point of landfall and hence subject to drastic intensity and structural changes as a result of interaction with land mass. The resultant development (or non-development) of rainbands and their attendant squalls also has a knock-on effect on the perception of “gustiness” or “storminess”. Both TC LAPS and SWIRLS are tools that rely

518

heavily upon short-term extrapolation techniques and as such do not have the capacity or capability to allow for such significant intensity-related changes both spatially and temporally. The lack of reliable objective guidance can sometimes make decision-making at the stage of warning cancellation rather tricky. The challenge remains to: (a) develop better NWP models in the simulation of landfalling TCs; and (b) incorporate such numerical information into the nowcasting or other expert systems for critical decision-making processes in the operation of warnings. 5.1.5.12 The automatic weather station(AWS) network, updated every minute, is very efficient to fix the center position of tropical cyclone moving over the land after landfall, while the radar and satellite image provide only very limited information on its structure. The position and intensity of Tropical cyclone EWINIAR (0603) is determined by the wind circulation and pressure measured at surrounding AWSs at KMA. The Forecast Analysis System (FAS), visualization tools based on interactive workstation at KMA, provides information on the movement of radar echoes along with the synoptic wind circulation overlaid with satellite images and spots of lightening. The System of Convection Analysis and Nowcasting (SCAN), adopted from U.S.A. on workstations at KMA, give guidance on realtime movement vector and intensity of storm cells associated with spiral band of tropical cyclone or rainband interacting with remnants of a tropical cyclone. Summary 5.1.5.13 Satellite observations are major source for the determination of position, radius, deepness of a tropical cyclone. Automatic Dvorak techniques are tested in many operational centers, and microwave channel data are increasingly used. 5.1.5.14 Many centers may use nowcasting tools for the warning of heavy precipitation and strong wind for landfalling tropical cyclone, for instance, Hong Kong observatory effectively use the tools on operational basis. 5.1.5.15 Various diagnostics and satellite observations are available through Internet, and more training and education is requested for forecasters to apply them. Recommendation 5.1.5.16 NMHSs should enhance their forecasting capabilities by utilizing current and emerging satellite data. 5.1.5.17 Dvorak-type tropical cyclone intensity estimates need to be validated by special in situ observation field campaigns in all basins that lack routine reconnaissance programmes. (Elsberry and Velden, 2003) 5.1.5.18 Training seminars and workshops are required to demonstrate on the application of AODT and other interpretation techniques of satellite observations. 5.1.5.19 It is desired to make a survey on the nowcasting tools of heavy rain and gust wind for the landfalling tropical cyclones to share them among forecasters 5.1.6 Warning presentation 5.1.6.1 The accuracy of the forecast as well as landfall still remain worrisome on the part of disaster response community as well as the ultimate user, the affected common people. In united States, the warning zones have historically averaged about 300 nm (556km) in length. This distance has evolved as a trade-off between the desire to provide maximum lead time and the necessity for keeping the size of the warning area within reasonable limits. In as much as a typical damage swath from a hurricane is

519

100 nm or less, approximately two-third of the residents within the coastal warning zone are over warned. One reason the length of the hurricane warning zone has not shrunk more is that the large increase in coastal population and property valuation has dramatically increased the vulnerability of the area and, thus, the consequences of a bad forecast. (OFC, 1997) 5.1.6.2 A major problem of issuing severe weather warnings is that the probability of the most extreme weather events occurring is very low; this means that warnings of severe weather that could cause the most damage are not always taken as seriously as they should be because they appear to be very unlikely. Therefore decision-making for events of this severity should be related not just to the probability of occurrence but also to the impact of the event. A simple example is to ask the question: if there was a 5% chance of rain would you play golf? Whereas if there was a 5% chance that the flight you had booked would crash, would you board the flight? The impact clearly is important to the decision 5.1.6.3 Information on the uncertainties in tropical cyclone forecast and intensity were conveyed to the public on the tropical cyclone forecast track map and other methods so as to manage their expectation on the forecast accuracy. 5.1.6.4 In Malaysia since Tropical Storm Greg, the TC warning has been revised to include direct TC impact. 5.1.6.5 The set of precautionary announcements and advisories based on pre-agreed courses of actions with the parties concerned were selected at the time to reflect prevailing circumstances to give maximum protection to the public. 5.1.6.6 In the USA, studies following hurricanes Camille and Eloise revealed that people who knew the difference between a “watch” and a “warning” were no more or less likely than others to evacuate. What does matter is that people are told explicitly what actions they need to take to protect themselves and why (WMO, 2002). 5.1.6.7 The S-S Scale and STiCKs - In general, tropical cyclone disaster scales serve two purposes: 1) to allow people to relate a specific intensity (wind speed) value to particular levels of damage or destruction so that they can make appropriate decisions; and 2) to help assess tropical cyclone intensity at locations where wind measuring devices have failed or been destroyed, or where they were non-existent. Here we discuss two similar Scales that were developed for different areas: the Saffir-Simpson Hurricane Scale (SSHS) (Saffir 1972, Simpson 1974) and the Saffir-Simpson Tropical Cyclone Scale (STiCkS) (Guard and Lander 1999). Sheifer and Ellis (1986) presented a formalized methodology for tropical cyclone damage assessment, which was incorporated into the latter Scale. In recognition of the utility and success of STiCkS, the French countries of the Southwest Pacific and the Southeast Indian oceans requested that the WMO translate the Scale into French, and in 2005, a French translation was produced (N. Lomarda, personal communication). While the original northern hemisphere version is being used in most of the southern hemisphere tropical islands, it is desirable to produce a southern hemisphere version in the near future. Other Scales have been developed in Australia and the Philippines (Amadore 1985). There are likely others. In operational practice, emergency managers should not use these Scales for real-time decisions, but instead should use the products provided by the National Meteorological Center or Tropical Cyclone Warning Center. However, the data in these Scales provide a good first guess for planning and response when better data are not available. 5.1.6.8 Perhaps it is time to conduct a survey and assessment of the existing Scales. The major strength of these scales is their simplicity and ease of understanding, especially with less sophisticated users. The scales can also be fine-tuned for a specific area or location. The major weakness with these scales is with the storm surge values. Storm surge can vary greatly depending on the storm size, speed of motion, prior intensity, and several other factors. Thus, the storm surge values can sometimes be one Category in error (B. Harper, personal communication).

520

5.1.6.9 The Australian scale - while also 5 category levels - is quite different to the US S-S scale. The scale is based on maximum wind gusts (3sec) using the theory and evidence that these strong gusts that are responsible for the most significant damage during an event. It is recognized that other factors will have a significant effect on the amount of damage - for example the length of exposure to these cyclonic conditions can have a significant effect on impact - particularly with strong cyclones where damage debris can accumulate - escalating the damage. The Australian scale covers tropical cyclones (T.Storms) and severe tropical cyclones (Hurricanes/Typhoons), whereas the S S scale is only for Hurricanes http://www.bom.gov.au/catalogue/warnings/WarningsInformation_TC_Ed.shtml 5.1.6.10 The Philippines scale - there are four public storm signals for purpose of categorizing the impacts of tropical cyclones. Delineating the areas under each warning signal depends on the forecast wind strength. Values within the range may be easily decided on, but values near the boundaries of each range may prove to be difficult and could result to over warning or under warning as the case maybe, which can result to post warning criticisms – from the media and the public at large- at the very least or productivity losses for over warning cases and damages and casualties in the under warning cases. Based on damages information, most impacts were caused by tropical depressions and severe tropical storms and at times, their interaction with the monsoon systems. From experience also, typhoons with high winds were mostly fast moving; since the Philippines are archipelagos, it only takes short time for typhoons to move across the land. It is the slow moving depressions and storms laden with lots of rainfall, which cause so much troubles for the country. 5.1.6.11 The distorted common perception of cyclone as a wind related phenomena is a barrier to understanding the danger related to storm surge and flooding. This also deters timely evacuation of people from critical areas. In many a case, the warning has maritime bias. The example of Bangladesh may be cited here. In Bangladesh, the cyclone warning issued by Bangladesh Department of Meteorology (BMD). The BMD issues cyclone warnings based on sea ports and river ports, which are in turn used for disaster preparedness. As can be seen from the contents of the warnings, in the existing warning system most of the signals depend on the cyclone track relative to a port. The numbering of the signals is confusing. The understanding is that higher is the number of signal the more is the intensity of the cyclone. But that is not true, in the existing system higher number of signals do not necessarily indicate stronger cyclone. For example, there is no difference among the danger signals (e.g. signal no. 5, 6 and 7) from the point of view of the intensity of cyclone. Signal 5 is hoisted when the cyclone is expected to cross the coast to the north of the port, whereas signal 7 is issued when it is expected to cross near or over the port. Signal no. 8, 9 and 10 (great danger signals) are issued when it is expected that port will experience severe weather from a storm of great intensity (normally which is 90 km/hour or more). But there remains a confusing point regarding maximum limit or the signals of great danger. 5.1.6.12 While several Members use TC warnings, several Members use separate warnings concurrently to warn the public of heavy rain, strong wind, storm surge, and high wave associated with tropical cyclones or their remnants as well as rain-induced disasters such as floods and landslides. 5.1.6.13 During the passage of tropical cyclones, presentations of tropical cyclone warning information and advisories in caring wording with a human touch by experts from NMHSs through live broadcasts on radio and TV will be effective in capturing public attention and relaying the latest critical weather information and advice. The media are interested in personalities. 5.1.6.14 Presentation and content are important, but there is no guarantee that people will heed even “perfect” warnings. Despite very accurate forecasts and warnings and round-the-clock television coverage for Hurricane Katrina, nearly 1300 people perished. A Special Report on the aftermath of Katrina summed up the challenge as: “The meteorological community once again needs to determine how to get better results out of its best efforts to fight the unthinkable” (Rosenfeld, 2005). How do we

521

battle human nature? We must better understand what motivates people to heed or ignore warnings. We must figure out how to convey probabilistic information in an easily understood manner. 5.1.6.15 Getting the science right is only ½ the equation. Hurricane Katrina has clearly demonstrated that good science is not enough to avert a disaster. There is a need to ensure that forecasters are trained in speaking to the media and that they develop credibility in vulnerable communities so that their message will be clearly heard, understood, and acted upon. In Australia, this is achieved through media skills training, on-going media exposure in at-risk communities and a policy of having the tropical cyclone forecaster do all radio interviews/broadcasts to affected communities (RAI Tropical Cyclone Committee, 2005). 5.1.6.16 Following the landfall of Hurricane Juan (2003) as SS2 storm in Nova Scotia, Canada, the Canadian Hurricane Centre (CHC) instituted the use of tropical storm and hurricane watches and warnings. These are issued by the CHC and are independent of elemental watches and warnings issued by the various Canadian Storm Prediction Centres and Forecast Offices. They are not just coastal bulletins, as in the United States, but extend inland through forecast areas. Summary 5.1.6.17 The uncertainty of tropical cyclone track and intensity is conveyed to the users in terms of probabilistic expression. It depends on the accuracy of forecasts and on the vulnerability of the target area. 5.1.6.18 The warning has to be interpreted in terms of action words to allow the users to respond on. The various scales on intensity have been developed, depending on the socio- economic conditions. Recommendation 5.1.6.19 Issue probabilistic forecasts of tropical cyclone/severe weather conditions up to 5 days ahead in all regions, to allow appropriate response. A better way is needed to communicate with the public so as to improve their understanding of probabilistic prediction. One solution for more understandable probabilistic prediction might be to translate it into a more illuminative language, for different stakeholders. Assistance from social scientists may be necessary in the translation process; 5.1.6.20 Conduct research to find out what people understand and do not understand, then evaluate through demonstration projects. 5.1.6.21 The warning message should contain information on what is to be done to minimize loss of lives and properties by the individuals and community. Such information will provide outlet for positive action on the face of cyclone threat. 5.1.6.22 good tropical cyclone warning system should be simple, easy to understand, and able to trigger organized responses of the government and orderly collective responses of the public to minimize loss of lives and damage to property. 5.1.6.23 Tropical cyclone warning symbols/categories should be considered as an efficient way of conveying tropical cyclone warning message to the public and facilitates the public triggering of a timely collective response, hence promoting effectiveness of the tropical cyclone warning. 5.1.6.24 Different forms of presentation of tropical cyclone messages should be developed and fitted to the strength of the dissemination channels and the level of the intended audience to enable effective communication of the warning to different sectors of the community. 5.1.6.25 Warning signals in the form of flags or similar symbols culturally acceptable to the community

522

should be adopted for easy dissemination of warning signals in societies with low literacy level and with limited communication infra-structure. 5.1.7 Warning dissemination 5.1.7.1 NMHSs need to be aware that the warning message may change in the transmission process and the recipients may not understand the warning and know how to respond in their own best interests. 5.1.7.2 In beach areas of south-east Florida during hurricane Andrew in 1992, 30 per cent of the residents said they did not hear from officials that they needed to evacuate their homes for safety. The influence of hearing vs. not hearing evacuation notices is greatest not in beach areas but in mainland locations that would flood dangerously only in strong storms (WMO, 2002). 5.1.7.3 To facilitate broadcast of the change in the tropical cyclone warning status by radio and TV stations simultaneously at the scheduled time and to avoid possible confusion by the public, the warning can be issued to the station 5 to 10 minutes earlier with an embargo on its release prior to the scheduled time. 5.1.7.4 In tourist cities, tourists and visitors will get to know tropical cyclone warning information from hotels and other sources. 5.1.7.5 Today, a majority of people appears to receive most of their warning information by television or radio, as has been documented by survey research with the public in Australia, Bangladesh and the USA. A great deal of hurricane information is accessible via the Internet but, as yet, fewer than 10 per cent of US households say they rely a great deal on that source for information about a threatening storm. (WMO, 2002). 5.1.7.6 Recent time has seen revolution in telecommunication technology. Now cyclone warning services have access to mobile telephones, satellite phone, satellite based warning broadcast system in addition to landline telephone networks. As emphasized earlier, public telephone network (landline/cellular) are effective before the disaster strikes. Once the disaster has struck, these networks are prone to failure/outages due to heavy traffic congestion, power breakdown etc. As such, these are not effective in dissemination of updated information to Disaster managers, broadcasters, affected communities etc. The role of electronic media (radio/TV) in informing the communities becomes limited because the broadcaster may not get latest warnings from warning service due to communication failure and the target audience may not be able to operate radio/TV due to power breakdown. In such situations, satellite based communications systems like satellite phone, VSAT terminal, Digital data/voice broadcast system, are much more effective in warning dissemination, as these are less prone to failure during adverse weather conditions. A satellite based voice broadcast systems known as “Cyclone Warning Dissemination System (CWDS)” has been functioning at several cyclone prone locations on the east and west coast of India, for the last several years and has proved quite effective in cyclone warning dissemination directly to people at risk. The system is owned and operated by India Meteorological Department(IMD). Cyclone Warning Centre prepares customized warning messages for each receiving location in local language and sends to specific location through satellite uplink. At the receiving end, when system is activated by Cyclone Warning Centre, the receiver sets off a loud siren to alert staff attached to receiver so that he can arrange to receive the warning message. The siren is followed by voice message containing details of adverse weather conditions likely to affect that area in local language. The message is noted by the staff and disseminated to local communities at risk. 5.1.7.7 KMA recently disseminates warning messages to digital multimedia broadcasting (DMB), and to the Internet portal sites.

523

5.1.7.8 In some countries like Cambodia, the GTS communication network is not available and the only means of communication with foreign countries is the Internet. 5.1.7.9 In developing societies, where most death occurs due to cyclone disaster, warning dissemination to the target population is challenging in view of poor infrastructure, limited access of population to mass media, remote location and the like factors. In such societies, a combination of modern and traditional media is to be used for warning dissemination. Community organizations and volunteers can be useful in disseminating warning in most developing countries. Throughout the warning phase of a cyclone, the CPP volunteers caution the people through megaphone and house-to-house contact as well as shift the endangered people to safe places and cyclone shelters. The headquarters of CPP maintained round the clock contact with the coastal region through 56 field wireless stations before, during and after the disaster. 5.1.7.10 Baker reports that the public is most effectively informed when authorities go door-to-door notifying residents of the need to evacuate or drive through neighborhoods announcing the evacuation over loudspeakers. (Schmidlin, 2006) 5.1.7.11 In some of the more developed countries for those who can afford it, the private sector has moved in to fill the shortfalls of the NMHSs. They specially tailor products and services for specific customers or sectors, but for a price. In some developed countries, the NMHSs themselves provide this service for a fee. 5.1.7.12 Automatic telephone inquiry system (telephone recorders and poll faxes) provides a direct and simple way for the public, especially the underprivileged who will have to rely on the phone, for acquiring the latest weather forecast and warning information. 5.1.7.13 Radio offers an extremely effective and widely adopted means of disseminating weather information and will continue to be one of the most common and critical components of the dissemination system of warning messages. 5.1.7.14 The effectiveness of the tropical cyclone warning system can be further enhanced by harnessing the power of the Internet. 5.1.7.15 The WMO websites for Severe Weather Information Centre (SWIC) and World Weather Information Service (WWIS) containing official weather forecasts and tropical cyclone warnings in different regions of the world demonstrated the synergetic effect of contributing NMHSs, maximizing the effectiveness of warnings at the global level. 5.1.7.16 Tropical cyclone products and information are disseminated through multiple channels in most Member countries such as amateur radio, HF radio, GTS, fax, US NWS network, Internet, IRC chat and voice, military communication circuits in JTWC, EMWIN, and RANET. Emergency Management Weather Information Network (EMWIN) and the RANET digital satellite radio broadcast are inexpensive ways to send communications to small, rural areas. 5.1.7.17 In Canada many media outlets have a reduced capacity on weekends, and in the case of some commercial radio stations, they may even be fully automated. In such cases, contacting media personnel to notify them of tropical warnings has proven problematic and emergency preparedness organizations have been engaged to aid in this process. Summary 5.1.7.18 The communication channels for the dissemination of warnings continue to expand to DMB, Internet portal sites, mobile phones, satellite broadcasting, along with conventional means.

524

5.1.7.19 However, the door to door notification is yet the efficient way of communication to personalize the risk. Recommendation 5.1.7.20 Ensure dependable and effective dissemination of nowcasts, forecasts, advisories, watches and warnings in real-time to decision-makers including emergency managers, media, general public and other stakeholders in most countries and regions. 5.1.7.21 NMHSs should if possible disseminate warnings through multiple and diverse channels with varieties of high and low technology with backup capabilities to facilitate users to respond to the warning in a timely manner. 5.1.7.22 Members may consider adopting the use of colored or numbered codes/categories in a tropical cyclone warning system as an effective way in conveying tropical cyclone warning status to the public. 5.1.7.23 NMHSs should take advantage of the advances in communication means such as wireless broadband access, GPS, and GIS technologies to enhance the relevance and effectiveness of warning. 5.1.7.24 Warnings should be disseminated through as many ways as possible such as :

(a) Text warning messages could be disseminated proactively to users via emails. (b) Dedicated web pages with audio warning alerts can be developed to provide tailor-made

weather services to special clients. (c) Web pages in a variety of formats, viz. graphics, text-only, audio, Wireless Application Protocol

(WAP) and Personal Digital Assistant (PDA), extensible Markup Language (XML) versions can satisfy the needs of diverse users.

5.1.7.25 Attempt should be made to develop volunteers at the community level under appropriate organizational arrangement as vehicle for tropical cyclone warning message dissemination on a door to door basis in a cost-effective manner. Such organization could also work on preparedness and response with community involvement. The Cyclone Preparedness Program (CPP) in Bangladesh can serve as an useful model of grassroots level organization for cyclone mitigation. 5.1.8 Education, promotion, and nonstructural means 5.1.8.1 More accurate and timely forecasts are also important for decrease of natural disasters, however, improvements in warning systems and in disaster management remain critical to mitigating the loss of lives and, to some extent, the damage over the world. Political and administrative decision makers are responsible for natural causes. They have to take these realities into account, not just in developing a vigilant disaster management system, but also in land-use planning, development of coastal districts, and insurance measures (Reason and Keibel, 2004). 5.1.8.2 One of the biggest obstacles to vulnerability reduction is the lack of public awareness of the threat and basic mitigation and preparedness measures. 5.1.8.3 Even when information is made public, it does not always reach the target audience. Surveys conducted by researchers at James Cook University in Australia following cyclones Justin and Gillian found that the majority of respondents had been unaware of cyclone safety information in their telephone directories. But general knowledge about tropical cyclones and associated safety rules is not sufficient in many cases. In the USA, people who could name hurricane safety rules, knew the definition of “low-lying area” and knew the difference between hurricane “watches and warnings” were no more

525

likely than others to evacuate in Camille and Eloise (WMO, 2002). 5.1.8.4 Interviews were conducted with residents of two communities in Bangladesh following the 1991 cyclone that killed 139 000 people. Most people said they did not believe the warnings or they were afraid of looting if they left their home unattended. The expression “it won’t happen here” is related to the failure to believe warnings. Haque indicated that warnings were not believed partly due to past false alarms. The notion “it won’t happen here” stemmed from the fact that the area had not been struck by a major storm since 1960. (WMO, 2002) 5.1.8.5 A great deal of research has been conducted in the USA to explain why some people evacuate and others do not. the examples of hearing evacuation notices, perception of vulnerability, and housing type have already been mentioned. In fact, those variables, plus the actual physical vulnerability of a person’s location (e.g. living on the beach susceptible to wave action and inundation even in weak storms vs. living on the mainland in areas only flooded by strong storms) account for the great majority of the variation in whether people evacuate in the USA. (WMO, 2002) 5.1.8.6 Familiarity of the tropical cyclone warning system by the public is essential for an effective and orderly implementation of response actions. 5.1.8.7 The best way to accomplish this is to incorporate the information into the curriculum of primary and secondary education systems. The Weather Forecast Office at Guam worked with middle school science teachers to develop a curriculum that addressed the natural hazards that affect and influences the lives of the people of the region. Some 45 hours of instruction and a multitude of materials were provided to the teachers. They developed the curriculum within 6 months, but it has yet to be reviewed and accepted by the public school system. There is tremendous competition for the hours that will be needed for the lessons to be taught and tremendous inertia to change. 5.1.8.8 Outreach can be an effective method to get specific information to a targeted audience. However, outreach requires a large investment of time for what is usually a relatively small audience. Thus, the payoff should be high. As an example, each year, the Weather Forecast Office in Guam conducts 20 workshops, makes 20 school presentations, entertains 25 school visits, addresses 15 civic groups, and talks to 220 different individuals about specific aspects of weather support. This amounts to about 2500 contacts per year for 300 events that last an average of 120 minutes. At the same time, the Office conducts an average of 50, 10-minute radio and television interviews per year, with an audience of 5000 people per interview. Which is best? It depends on the purpose. Outreach provides a lot of information to a small audience over several hours, while media passes a little information to a lot of people over a few minutes. Cottrell (2004) demonstrated the value of the local knowledge and local survival strategies, specifically of women in Northern Australia, in increasing the effectiveness of mitigation planning. 5.1.8.9 NMHSs have to be aware that warnings and hazard awareness need to be internalized by the society through time, cyclone warnings have to be recognized in time and space ahead of or alongside the hazard. Warnings and responses are at the opposite ends of a chain of processes. Failure of links between and among players in the warning process will reduce effectiveness. The players may include meteorologists/ hydrologists, government decision makers, emergency management disaster planners, media, local government units, community action and response teams as well as individuals. 5.1.8.10 In the Philippines, establishment of a community-based early warning system for floods, landslides and flashflood is being set-up in various communities around the major river basins in the country. The idea here is to empower local government units and community volunteers to develop localized indicators that will relate flashfloods for example, to rainfall observations. The success of such community-based early warning system relies heavily on the design of the flow of communications from the observation sites to the decision-maker and the timing of the warning to reach the affected community for them to evacuate to safe areas at the shortest time possible. Also, the sense of

526

ownership of the community for the early warning system contributes its effectiveness. The major role of the national hydrometeorological service is to be able to transfer the “technical know how” to volunteers in a pragmatic way. Both the NHMS and the community are expected to give their commitment, cooperation and support to make the system sustainable. This effort requires a lot of coordination with local government units, public information drive and genuine interest and faith in the success of the system. 5.1.8.11 Community-based approach is one of the social instruments being used to involve stakeholders in the decision-making process particularly the most disadvantaged or affected groups. It is part of a larger participatory process. In many developing countries including the Philippines, this approach is being employed mostly in environmental concerns. Four years ago, we included community-based flood early warning system (CBFEWS) in one of the activities in the Typhoon Committee. This is a response of TC to involve the three disciplines of meteorology, hydrology and disaster prevention in one activity. Malaysia, Thailand and the Philippines participated in a pre-workshop on this activity. 5.1.8.12 The Hong Kong Observatory and other local government departments jointly organized a year-long public education campaign in 2005 to promote public awareness and understanding of natural hazards. The campaign involved the active participation of the public (“Safer Living-Reducing Natural Disasters” campaign and the “Tropical Cyclone Name Nomination Contest”). The public supported this initiative. 5.1.8.13 Media liaison officers in NMHS are important in developing close connections among various stakeholders in the warning and response process. People have different types of attitudes and perceptions towards risk, warning and disaster preparedness education. It is important to promote the appropriate disaster preparedness education and school curriculum packages having in mind the intended audience. 5.1.8.14 In many countries, public awareness activities, whatever may be the scale, is mostly confined to tropical cyclone seasons. This should give way to year round public awareness activities on a sustained basis to create public understanding and consequent preparedness and mitigational behavior at the level of the community. 5.1.8.15 In most developing societies, coastal inhabitants are reluctant to evacuate or to move to cyclone shelters due to attachment to their worldly possessions including that of livestock and poultry. The impact study of Bangladesh May 1994 cyclone by Helen Killer International indicated that most residents in the affected area waited until the wind flow started to seek shelter or did not move until their home collapsed. In the event that the cyclone had been more severe, 5.1.8.16 A further challenge of ET for the forecaster is communicating with emergency management personnel and the public, particularly since there is much less public awareness of the hazards associated with ET than of those due to a tropical cyclone. (Jones et al., 2003) 5.1.8.17 In Canada, Hurricane Juan (2003) was a stark reminder of the realistic threats of hurricanes when, in spite of solid forecast guidance from the CHC, the public were found to be generally unprepared. A post-event assessment of failing points yielded a variety of reasons for the poor response:

a) A significant over-reaction of the media to Tropical Storm Isabel (9 days earlier) left the media feeling timid about repeating the false alarm. Accordingly, it was extremely difficult to engage them to the fullest extent required.

b) Since it had been more than 100 years since the city of Halifax had experienced a hurricane of that magnitude there was no corporate memory for people to draw from – they were essentially unschooled in the threats of a category 2 hurricane. Accordingly, most people didn’t know how to react.

527

c) Many had a false sense of security based on their experience with SS1 Hurricane Hortense (1996) . . . a storm which resulted in some (but minimal) impacts to Halifax since it made landfall east of the city. The public felt that that they knew what a hurricane could do and were unprepared for a stronger hurricane making landfall just west of the city.

d) Many/most residents of Nova Scotia believed that they were not seriously threatened by hurricanes, or that impacts, if any, would be insignificant (Hanson, 2003).

e) Of the few that did believe that hurricanes could strike Nova Scotia, most felt that the cold water surrounding the province would mitigate the effects by weakening the storm. At the time of Juan, however, SSTs were running 3-5 degrees C warmer than normal . . . a fact that resulted in Juan remaining at category 2 strength (Fogarty et al., 2006).

f) Much of the general public had a poor sense of vulnerability. g) Many believed that since they had been impacted by Hortense in 1996 that Juan would NOT

strike because hurricanes do not occur with that frequency. h) Of the few that paid attention to the track forecasts (disseminated through the internet and the

media), a great percentage placed too great an emphasis on the track-line and gave too little credence to the worded statements that talked about impacts extending a significant distance from the track line.

i) Many felt that although the CHC statements were accurate and factual, they lacked the necessary dire-nature to engage the public at the highest level.

5.1.8.18 Even in the presence of structural means, these is a necessity for comprehension of flood disaster prevention and mitigation through inclusion of nonstructural means including:

(a) Legal framework, coordination among stakeholders (b) Land use (spatial) management (c) flood monitoring, forecasting & early warning systems (d) Preparedness –hazard mapping, improving communication, education to raise awareness (e) Insurance and mutual aid

Summary 5.1.8.19 The public does not always respond to protect from the danger of tropical cyclone even when the warning itself is accurate and disseminated in time. Numerous factors influence the behavior of users including the false alarms, rare occurrence, belief “it won’t happen here”, and so on. 5.1.8.20 Various activities are on-going to enhance public awareness on the risk of tropical cyclone through outreach programs, primary and secondary education, and cooperation among decision makers, emergency managers, media, stake holders at community level. Recommendation 5.1.8.21 Increase investment in awareness programmes related to the risks and consequences of natural hazards for decision-makers, emergency managers, media, NGOs, public and other stakeholders for prompt and effective response at the national to community levels; 5.1.8.22 Educate stakeholders annually on proper interpretation of forecasts, advisories, warnings and other meteorological and hydrological information. The goal is to conduct at least one session for each stakeholder; and 5.1.8.23 NMHSs should collaborate with scientists and researchers to develop TC disaster scenarios and visualize hazards in the form of hazard map, risk map or disaster management map, designating evacuation sites, and displaying warning/evacuation signs etc. All stakeholders including national and local governments, non-government organizations, private sectors and individuals should develop disaster management plans stipulating the roles of different stakeholders. The plans should be updated constantly to reflect the current situation. Drills of disaster management plans and test of

528

communication links involving all stakeholders should be organized by local government. 5.1.8.24 DPP should gear towards the prediction of extreme weather events. Studies on the climate variability and change impacts on extreme weather events should be carried out to enhance disaster preparedness. 5.1.8.25 The tropical cyclone warning criteria and protective measures must be publicized through as many channels as possible such as websites, publications for teaching materials, guidelines, demonstration of past achievements on the assessment of vulnerability and risk management distributed at schools, district offices, entry ports, resource centers, short advertisement broadcast by radio and TV. 5.1.8.26 Promote public education and conduct outreach activities such as talks, exhibitions, school visits, TV documentary series, education and publicity campaigns to raise awareness and people’s preparedness for possible tropical cyclone related disasters, community empowerment and awareness. 5.1.8.27 Include materials on tropical cyclone warning system, tropical cyclone related hazards and corresponding disaster preparedness in the school curriculum and educational resources to promote public education to students and teachers. 5.1.8.28 NMHSs must formulate strategies on disaster risk management. These include (a) establishing a legal frame that will allow coordination mechanisms and protocols; (b) integrating disaster reduction concepts into land use planning; (c) improving information sharing and management; (d) promoting education and public awareness for example by integrating disaster education into the curriculum; (a) developing multi-stakeholder partnerships and public participation. 5.1.8.29 Knowledge, Attitude and Practice (KAP) survey on cyclone warning should be initiated on systematic basis in every tropical cyclone prone country on periodic basis and after every major cyclone to establish benchmarks and monitor progress in public understanding and internalization of cyclone warnings in the life style of individuals and communities inhabiting the coastal belts. Such survey findings would be very useful pointer to policy makers for investment decisions and NMHSs to assess preparedness of the communities at risk of losing invaluable lives. KAP survey on cyclone warning should be also conducted on other stakeholders viz political decision makers, disaster preparedness personnel, non-government organizations and others involved in DPP activities to identify the need for training and orientation type of activities to ensure provider preparedness. 5.1.8.30 Social impact study of tropical cyclone should be conducted on periodic basis with joint team of meteorologists, hydrologists and social scientists in every cyclone prone society in line with the recommendation of IWTC-V. A compilation of such studies should be published to promote the ‘culture of promotion’ as advocated by WMO 5.1.8.31 The accuracy of warning can be directly evaluated through a post event verification scheme, comparing the forecasts included in the warning against the observed. Field surveys and damage assessment may also help establish the degree of correctness of the delineation of areas under a certain warning category. 5.1.8.32 In societies with dense population, the pressure on land plus attraction of coastal land fertility attracts people to settle in fringe land or islands in the coast. Regulatory framework on land use in the coastal areas should be developed and enforced to reduce vulnerability.

529

5.1.9 International cooperation 5.1.9.1 Larger warning centers may use specially designed computer systems that manage and co-register numerous sources of data and computer model output, and allow the forecaster to formulate the forecast into pre-formatted bulletins and products. These products may be sent over a wide variety of dissemination systems. Other NMHSs may have to develop the warnings from a limited amount of data and computer model guidance from a few disjointed sources, and then manually produce the warning on a simple word processing system. These products may be sent to a very limited number of dissemination systems. 5.1.9.2 Almost under-developing countries do not have numerical weather prediction (NWP) capacity, and the challenges and potential solutions for improved tropical cyclone forecasting for these regions are required. In order to improve of NWP over these countries, international cooperation is indispensable. Moreover, national forecasters must at least be in a state of heightened alert of tropical cyclones, in advance by sharing the information through world wide network. The international training programs for typhoon forecasts are able to contribute advance to other countries (Raghavan and Rajesh, 2002). 5.1.9.3 Various EPS products are available on the Web, and will increase in the future through international projects under THORPEX Interactive Gland Global Ensemble (TIGGE). 5.1.9.4 The EPS model products were increasingly becoming useful for tropical cyclone forecasting. It was, however, noted that most countries of the SWIO were not yet familiar with the use of these products for tropical cyclone forecasting. (RAI Tropical Cyclone Committee, 2005) 5.1.9.5 A request was made to the JMA that real-time track forecasts for tropical cyclones be provided for the Southern Hemisphere, particularly the South Pacific Ocean. This would be a valuable addition to the limited suite of NWP-based TC track products available to the SW Pacific Nations that can be used for more effective ensemble-type forecasting techniques. (RAV Tropical Cyclone Committee, 2004) 5.1.9.6 The effectiveness of tropical cyclone warning can be enhanced by inter-agency partnerships and cooperation. In the process, effective communication and interaction through capacity building, governance, community empowerment, and disaster preparedness program are critical. 5.1.9.7 Cambodia, Vietnam, the Philippines, DPR Korea, and Lao PDR expressed the need for additional rain gauges to monitor rainfall for improving flood forecasts. Summary 5.1.9.8 The resource and information is limited in the developing countries to provide warning service for a tropical cyclone Recommendation 5.1.9.9 NMHSs should foster closer regional coordination and collaboration in the application of model and other forecasting guidance in tropical cyclone prediction and enhancement of telecommunications capability. Local NMHS can then adjust the service to cater for local characteristics.

530

Bibliography Aberson, S. D. and C. R. Sampson, 2003: On the Predictability of Tropical Cyclone Tracks in the Northwest Pacific Basin. Mon. Wea. Rev., 131, 1491–1497. Amadore, L. A., et al., 1985: Preliminary typhoon damage scale in the Philippines, PAGASA, Quezon City, Philippines. Cottrell, A, 2004: Incorporating local knowledge in mitigation for hazardous weather. Abstracts from the International Conference on Storms, Mecure Hotel, Brisbane, Australia, 5-9 July 2004. pp 116-117. Elsberry, R. L., and C. Velden, 2003: A survey of tropical cyclone forecast centers – uses and needs of satellite data. WMO Bulletin, July, 1-7. Emanuel, K., 2005: Increasing destructiveness of tropical cyclones over the past 30 years. Nature, 436, 686-688. Fogarty, C. T., R. J. Greatbatch and H. Ritchie, 2006: The Role of Anomalously Warm Sea Surface Temperatures on the Intensity of Hurricane Juan (2003) during Its Approach to Nova Scotia. Mon. Wea. Rev., 134, 1484–1504. Guard, C. P., and M. A. Lander, 1999: A scale relating tropical cyclone wind speed to potential damage for the tropical Pacific Ocean region: A user’s manual. WERI Technical Report 86, University of Guam, Mangilao, Guam, 60 pp. Hanson, R., 2003: Actual Versus Perceived Risk Due to Hurricanes in Nova Scotia. Department of Geography Honors Thesis, unpublished honors thesis, Saint Mary's University, Halifax, NS. Hart, R.E., 2003: A Cyclone Phase Space Derived from Thermal Wind and Thermal Asymmetry. Mon. Wea. Rev., 131, 585–616. Jones, S. C., P. A. Harr, J. Abraham, L. F. Bosart, P. J. Bowyer, J. L. Evans, D. E. Hanley, B. N. Hanstrum, R. E. Hart, F. Lalaurette, M. R. Sinclair, R. K. Smith and C. Thorncroft, 2003: The Extratropical Transition of Tropical Cyclones: Forecast Challenges, Current Understanding, and Future Directions. Weather and Forecasting, 18, 1052–1092. Klotzbach, P. J., and W. M. Gray, 2003: Forecasting September Atlantic basin tropical cyclones activity. Weather and Forecasting, 18, 1109-1128. Knaff, J. A. , C. R. Sampson and M. DeMaria, 2005: An Operational Statistical Typhoon Intensity Prediction Scheme for the Western North Pacific. Weather and Forecasting, 20, 688–699. Krishnamurti, T.N., C.M. Kishtawal, T.E. LaRow, D.R. Bachiochi, Z. Zhang, E. Williford, S. Gadgil and S. Surendran, 1999: Improved weather and seasonal climate forecasts from multimodel superensemble. Science, 285, 1548-1550. Kumar, T. S. V. V. , T. N. Krishnamurti, Michael Fiorino and Masashi Nagata, 2003: Multimodel Superensemble Forecasting of Tropical Cyclones in the Pacific. Mon. Wea. Rev., 131, 574–583. Negri, A. J. , N. Burkardt, J. H. Golden, J. B. Halverson, G. J. Huffman, M. C. Larsen, J. A. McGinley, R. G. Updike, J. P. Verdin and G. F. Wieczorek, 2005: The Hurricane–Flood–Landslide Continuum. Bull. Amer. Meteorol. Soc., 86, 1241–1247.

531

OFC, 1997: NATIONAL PLAN FOR TROPICAL CYCLONE RESEARCH AND RECONNAISSANCE 1997-2002. FCM-P25-1997, Office of the Federal Coordinator. RAI Tropical Cyclone Committee, 2005: Final report 37th session of Tropical Cyclone Committee for the Southwest Indian Ocean, 3-7 October 2005, Garborone, Boswana, 22pp. RAV Tropical Cyclone Committee, 2004: Final report 10th session of Tropical Cyclone Committee for the South Pacific and southeast Indian-Ocean, 10-15 July 2004, Brisbane, Australia, 17pp. Raghavan, S. and S. Rajesh, 2002: Trends in Tropical Cyclone Impact: A Study in Andhra Pradesh, India. Bull. Amer. Meteorol. Soc., 84, 635–644. Reason, C. J. C. and A. Keibel, 2004: Tropical Cyclone Eline and Its Unusual Penetration and Impacts over the Southern African Mainland. Weather and Forecasting, 19, 789–805. Rosenfeld, J., et al, 2005: The Mourning after Katrina. Bull. Amer. Meteorol. Soc, 86, 1555-1566. RSMC Tokyo Typhoon Center, 2005: ACTIVITIES OF THE RSMC TOKYO-TYPHOON CENTER IN 2005, working document for 38th session of Typhoon Committee (14-19 Nov., 2005, Hanoi Vietnam) Saffir, H. S., 1972: The nature and extent of structural damage caused by Hurricane Camille, National Oceanic and Atmospheric Administration, Washington, DC Schmidlin, T. W., 2006: On evacuation and deaths from hurricane Katrina. Bull. Amer. Meteorl. Soc., June, 754-756. Sheifer, I. C., and J. O. Ellis, 1986: Development of a tropical cyclone damage assessment methodology. NOAA Technical Memorandum NESDIS AISC 4, 42 pp. Simpson, R. H., 1974: The hurricane disaster potential scale. Weatherwise, 27, 169-186. Trenberth, K. E., 2005: Uncertainty in Hurricanes and global warming. Science, 308, 1753-1754. Typhoon Research Coordination Group (TRCG), 2005: Final report of the Workshop on Effective Tropical Cyclone Warning, Shanghai, China, 24-28 April 2005, Webster, P. J., G. J. Holland, J. A. Curry, H. R. Chang, 2005: Changes in tropical cyclone number, duration, and intensity in a warming environment. Science, 309, 1844-1846. World Meteorological Organization, 2002: The science and forecasting of tropical cyclones. WMO/TD-No. 1129, 95 pp.

532

Acronym AMSU Advanced Microwave Sounding Unit AODT Advanced Objective Dvorak Technique AWS Automatic Weather Station BMD Bangladesh Department of Meteorology CBFEWS Community-based Flood Early Warning System CHC Canadian Hurricane Centre CIMSS Cooperative Institute for Meteorological satellite Studies CLIPER CLImatology and PERsistence CPP Cyclone Preparedness Program CPS Cyclone Phase Space CWDS Cyclone Warning Dissemination System DMB Digital Multimedia Broadcasting DMO Direct Model Output EMWIN Emergency Management Weather Information Network ET Extratropical Transition FAS Forecast Analysis System GEOSS Global Earth Observing System of Systems GTS Global Telecommunication Network HKO Hong Kong Observatory IMD India Meteorological Department JMA Japan Meteorological Agency JTWC Joint Typhoon Warning Center KAP Knowledge Attitude and Practice KMA Korea Meteorological Administration LAPS Local Analysis and Prediction System MAE Mean Absolute Error NCEP National Center for Environmental Prediction NMHS national Meteorological and Hydrological Service NRL Naval Research Laboratory NWP Numerical Weather Prediction OFC Office of the Federal Coordinator PDA Personal Digital Assistant QPF Quantitative Precipitation Forecasting QuikScatQuick Scatterometer RANET RAdio and interNET RSMC Regional Specialized Meteorological Center SCAN System of Convection Analysis and Nowcasting SSHS Saffir-Simpson Hurricane Scale StiCKs Saffir-Simpson Tropical Cyclone Scale SWIC Severe Weather Information Centre SWIRLS Short-range Warning of Intense Rainstorms in Localized Systems TC Tropical Cyclone THORPEX The Hemispheric Observing system and Predictability EXperiment TIGGE THORPEX Interactive Gland Global Ensemble UKMO United Kingdom Meteorological Office USGS U.S. Geological Survey WMO World Meteorological Organization WWIS World Weather Information Service XML Extensible Markup Language

533

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES Topic 5.2 : Factors Contributing to Human and Economic Losses Rapporteur: Roger A. Pielke, Jr. Center for Science and Technology Policy Research University of Colorado 1333 Grandview Ave, Campus Box 488 Boulder, Colorado 80309-0488 Email : pielke@colorado.edu Working Group: Ryan Crompton, Eberhard Faust, Joel Gratz, Manuel Lonfat, Qian Ye, and S. Raghavan,

Disclaimer: The views expressed in this paper reflect those of the individual authors contributing, and do not necessarily reflect the views of any organizations with which each may be associated.

Outline

5.2.1 Understanding human and economic losses 5.2.2 Cost-benefit studies of disaster mitigation 5.2.3 Tropical cyclone case studies

a) India b) Australia c) United States

5.2.4 Differing views of the role of global warming on losses 5.2.1: Understanding human and economic losses Human losses refer to the loss of life and injuries that result from the impacts of a catastrophic natural event, such as a wind storm for example. Losses from individual tropical cyclones have been catastrophic across centuries. Rappaport and Fernandez-Partagas (1995, updated by Beven 1997) have for example compiled human loss data for the Atlantic 1492-1997 that shows human impacts on early Europeans travelers for example.6 The Center for Research on the Epidemiology of Disasters (CRED) also maintains a database of global disaster called EM-DAT (http://www.em-dat.net/index.htm).7 CRED urges caution in using disaster data:

Data on disaster occurrence, their effect upon people and cost to countries remain at best patchy. No single institution has taken on the role of prime provider of verified data. The data in

6 http://www.nhc.noaa.gov/pastdeadly.shtml 7 GUHA-SAPIR., D. HARGITT, D. HOYOIS, Ph. (2004). Thirthy years of natural disasters 1974-2003: The numbers, Presses Universitaires de Louvain: Louvain-la Neuve. http://www.em-dat.net/documents/Publication/publication_2004_emdat.pdf

534

EM-DAT is culled from a variety of public sources, including reports by governments, insurance companies, press agencies and aid agencies. The original information is not specifically gathered for statistical purposes and inevitably, even though CRED applies strict definitions for disaster events and parameters, the original suppliers of the information may not. The figures should be regarded as indicative.8

The U.S. National Hurricane Center maintains a database of economic losses related to hurricanes that impact the US and, to some extend, the Caribbean region and Central America (e.g., http://www.nhc.noaa.gov/Deadliest_Costliest.shtml). Several reinsurance companies also maintain global tropical cyclone loss databases (including Munich Re9 and Swiss Re10). Economic damage is often defined as the direct damages associated with a hurricane’s impacts as determined in the weeks (and sometimes months) after the event. Indirect damages and longer-term macro-economic effects, such as the surge in the price of materials and qualified workers after an event, are sometimes considered in a tabulation of losses. Inland flooding related to tropical cyclones is handled inconsistently across datasets. Different methods exist for calculating a disaster’s impacts and result in correspondingly different estimates for the same event. Consequently extreme caution should be taken when integrating analyses or conclusions across datasets. For these reasons, a major workshop held in May, 2006 in Hohenkammer, Germany organized to look at economic losses recommended the “creation of an open-source disaster database according to agreed upon standards.”11 12 While this recommendation has a broad focus, tropical cyclones would form a substantial part of any such database given the magnitude of their associated human and economic losses. Munich Re has begun discussions about the creation of such an open-source database, built upon their NatCat Service. We recommend that the tropical cyclone community should collaborate on this, or other, efforts to create a centralized, comprehensive, peer-reviewed database. In particular, significant by their absence in this report, data on the non-economic human losses related to tropical cyclones are not readily available. 5.2.2: Cost-benefit studies of disaster mitigation While there is general agreement among experts that reducing vulnerability to tropical cyclones through disaster mitigation is the most effective strategy for addressing human and economic losses, formal evaluation of specific options for reducing disaster vulnerability remains difficult because “natural hazards and related vulnerability are rarely considered in the design and appraisal of development projects. Similarly, monitoring and evaluation are still relatively neglected in disaster reduction, especially where impact evaluation is concerned.”13 There are a number of benefit-cost estimates that circulate in the disaster community that suggest that “disaster mitigations pays,” typically by a ratio of 3 to 1 or higher. Of such estimates made for disaster mitigation projects around the world, which

8http://www.em-dat.net/documents/Publication/publication_2004_emdat.pdf 9http://www.munichre.com/publications/30203901_en.pdf?rdm=511#search=%22munich%20re%20natcat%20service

%22 10http://www.swissre.com/INTERNET/pwsfilpr.nsf/vwFilebyIDKEYLu/EWAL6MBJQ2/$FILE/Sigma2_2006_e.pdf 11http://sciencepolicy.colorado.edu/sparc/research/projects/extreme_events/munich_workshop/workshop_report.pdf 12http://sciencepolicy.colorado.edu/sparc/research/projects/extreme_events/munich_workshop/workshop_report.pdf 13 C. Benson and J, Twigg, 2004. 'Measuring Mitigation': Methodologies for assessing natural hazard risks and the net benefits of mitigation - A scoping study, ProVention Consortium. Jhttp://www.proventionconsortium.org/themes/default/pdfs/MM_scoping_study.pdf

535

includes tropical cyclones as well as other types of disasters, Benson and Twigg (2004, pp. 13-14)14 write,

However, there is surprisingly little evidence in support of many broad-brush statements. Detailed underlying calculations are not available, suggesting that they may, in fact, be no more than ‘back-of the-envelope’ – if informed – estimates. Even if they are based on more extensive calculations, the fact that the workings underlying them are not readily available can cast doubts on their legitimacy, particularly if figures involve some valuation of non-tangibles. Of course, financial analysis of loss and the cost of investments needed to avoid loss may not be sufficient to ensure greater attention to natural hazard risk, as demonstrated from experience elsewhere (for instance, in relation to disease, water pollution and illiteracy). But proof of net financial benefits is almost undoubtedly a first, very necessary step in making a case for the importance of analysing hazard-related risks.

The lack of a well-developed body of disaster mitigation cost-benefit analyses of the value of disaster mitigation sets the stage for a chicken-and-egg problem. Because such studies are rare, it can be difficult to compare projects or policies focused on disaster mitigation with other sorts of development policies; hence disaster mitigation policies are at risk of being overlooked in any systematic comparison of costs and benefits across different policy alternatives. But if such projects are overlooked, then there is less incentive to call for and support rigorous cost-benefit studies. One consequence of this dynamic is that funds for disaster relief in the aftermath of a horrific disaster are, in many cases, easier to secure than funds for long-term reduction of vulnerability to disasters, which may have supported efforts that would have reduced the need for post-disaster relief. This vicious cycle is well-appreciated by observers of disaster policy, but remains entrenched.15 Thus, one recommendation is for increasing attention to the need for rigorous cost-benefit analyses of disaster mitigation policy alternatives and practices. To put this another way, irrespective how such studies turn out (in terms of relative costs and benefits) there is a substantial benefit to decision making related to disasters to be gained from a more comprehensive and rigorous understanding of the value of disaster mitigation. More fundamentally, there is also a pressing need for more thorough information on the broad human and economic impacts of disasters, 16 as well as indicators of relative vulnerability17 in order to help prioritize disaster mitigation investments. Even in the United States and Europe, where response to disasters is generally quite successful as measured by lives lost and community recovery from disasters there is little information available on the costs and benefits of disaster mitigation, as well as the role of government investments in disasters on outcomes.18 Actions to reduce the impacts of natural disasters are many and varied around the world. Some of these actions are developed in the longer-term, such as building codes, evacuation plans, and emergency response plans. Some of these longer-term plans require additional action in the

14 Ibid. 15 B. Wisner, P. Blaikie, T. Cannon, and I. Davis, 2004. At Risk - Natural hazards, people's vulnerability and disasters, Wiltshire: Routledge, London, UK. 16 For example, C. Benson and E. Clay, 2004, Understanding the Economic and Financial Impacts of Natural Disasters. Disaster Risk Management Series No.4. Washington DC: World Bank. http://www-wds.worldbank.org/servlet/WDS_IBank_Servlet?pcont=details&eid=000012009_20040420135752 17 UNDRO 1990, Preliminary Study on the Identification of Disaster-Prone Countries Based on Economic Impact. New York/Geneva: United Nations Disaster Relief Organization. 18 Meade, C. and Abbott, M.: 2003, Assessing Federal Research and Development for Hazard Loss Reduction, RAND, Santa Monica, CA.

536

short-term in the face of an impending threat, such as an order of evacuation. In some instances not only has a cost-benefit analysis not been attempted, but neither has a more general risk assessment. Hurricane evacuation in the United States is an example of such a situation. Hurricane forecasts of storm tracks have improved steadily over the past three decades or so, yet at the same time the area of coastline warned per storm increased for much of this time, then decreasing in the last four years. This suggests that decisions makers (including forecasters and emergency managers) have possibly become more risk averse over time and have used advances in the science of forecasting to reduce the chances of leaving part of the population unwarned. Of course, such strategies have costs, in the form of a greater number of people warned unnecessarily. But to date there has been little demand for the quantification of the costs, benefits, and risks associated with different approaches to the challenge of hurricane evacuation in the face of uncertainty.19 Arguably, the case of hurricane evacuation is representative of the broader challenges of evaluating existing disaster mitigation policies in terms of their costs, benefits, and risks. There are new and innovative policy options that have been proposed for disaster mitigation that will likely stimulate demand for greater attention to costs and benefits. Among these are the securitization of risk through financial products such as catastrophe bonds and derivatives20 and the provision of micro-finance21 in developing countries as a tool of disaster recovery in ways that reduce long-term vulnerabilities. Such policies are not widespread, however, and they have not been subject to rigorous evaluation of costs and benefits, but nonetheless have strong support among many disaster experts. Any effort to prioritize investments in improving preparation for and responses to tropical cyclone impacts – including the scientific and technological resources at the focus of the IWTC – would benefit from greater rigor in the collection of human and economic loss data, as well as research on the role that various policies, strategies, technologies, and other interventions have in shaping patterns of loses in particular locations. 5.2.3: Tropical Cyclone Case Studies The following three sections describe the economic losses related to tropical cyclones in India, Australia, and the United States. The omission of the northwest pacific basin in the following cases is a reflection more of the state of the literature in this area than that basin’s significance for tropical cyclone landfalls.

a) India (prepared by S. Raghavan)

Andhra Pradesh (location shown in Fig. 5.2.1) is one of the most tropical cyclone-prone states in India with a coastline of about 1030 km. Reliable meteorological and economic data are available from around 1970. In a study of tropical cyclones hitting the state in the period 1971-2000 Raghavan and Rajesh (2003)22 showed that there was no increasing trend in tropical cyclone frequency or intensity, (actually there was a negative trend during these three decades) but the estimated damage increased over the period. However after the damage was normalized using changes in inflation, increases in population and economic activity in the region, there was no corresponding trend in the normalized

19 Pielke, Jr., R. A., 1999: Hurricane Forecasting. Science, 284, 1123. 20http://www.axa.com/lib/axa/uploads/cpsocietes/2006/United_nations_PR_20060306.pdf 21http://www.proventionconsortium.org/themes/default/pdfs/microfin_guidebook.pdf 22 Raghavan S. and S. Rajesh, 2003, “Trends in tropical cyclone impact:: A study in Andhra Pradesh, India”, Bull. Amer. Meteor. Soc., 84, 635-644.

537

damage values. This is consistent with similar studies of U.S., Caribbean, and Latin American hurricane impacts (Pielke and Landsea, 199823; Pielke et al., 200324) and highlights the key factors contributing to increased losses. Losses are largely attributed to societal factors elsewhere as well 25 26 27 28. The study could not be extended to other coastal states in India because reliable damage data is not available. There is however no reason to expect very different results in those states. Further increases in damage is expected from future tropical cyclones. Funds directed by administrators towards disaster preparedness and post-disaster relief have also been examined. The figures for Andhra Pradesh (Source: Government of Andhra Pradesh) are as below.

Fig. 5.2.1 Locator Map of Andhra Pradesh, India.

23 Pielke, R. A. Jr., and C. W. Landsea, 1998, ”Normalized hurricane damages in the United States: 1925–95”, Wea Forecasting, 13, 621–631. 24 Pielke R.A. Jr., J. Rubiera; C. Landsea, M. L. Ferna´ndez, and R. Klein, 2003, “Hurricane Vulnerability in Latin America and The Caribbean: Normalized Damage and Loss Potentials”, Natural Hazards Review, 101-114. 25 Association of British Insurers (ABI) , 2005, “Financial risks of climate change”, Summary Report, 40 pp., www.abi.org.uk 26 R. Crompton, J. McAneney and R. Leigh, 2006, “Natural disaster losses and climate change: An Australian perspective”, Climate change and disaster losses workshop, Hohenkammer, Germany, 25-26, May 2006, pp 36-44. 27 Shi Jun, 2006, Climate change and disaster losses workshop, Hohenkammer, Germany, 25-26, May 2006, pp 118-120. 28http://sciencepolicy.colorado.edu/sparc/research/projects/extreme_events/munich_workshop/workshop_report.pdf

538

EXPENDITURE29 Indian Rupees Crore (Crore means 107 )

Relief following an event (includes tropical cyclones, floods and droughts), from 1979-80 to 1999-2000.

2781

Estimated damage due to a single tropical cyclone in 1996 6129

Expenditure on preparedness Not available

World Bank-aided Project (1990’s), mostly dedicated to infrastructure development

801

Apart from the amount contributed under the World Bank project, the level of expenditure allocated to preparedness was not available. This shows that much more importance is placed on relief after the event rather than proactive preparedness before it. This attitude is understandable in the sociopolitical context and is not confined to Andhra Pradesh. One only has to draw a parallel to the level of preparedness in New Orleans, U.S., prior to Hurricane Katrina, despite elaborate plans being available. Although in recent decades people in the region have become more aware of the need to be prepared for future tropical cyclones, there remains a “Fading Memory Syndrome” where the level of disaster preparedness diminishes in the absence of a recent event (of the order of a few years) impacting the area.30 The phenomenal damage and deaths resulting from the “super cyclone” (defined as an event with maximum sustained surface wind in excess of 62 m s-1) which struck the Indian State of Orissa with 72 m s-1 maximum sustained surface winds in October 1999, may be largely attributed to lack of preparedness.

b) Australia (prepared by R. Crompton)

The Insurance Council of Australia (ICA) Natural Disaster Event List is a comprehensive database of insured losses due to natural hazards. The Disaster List begins in 1967 and includes details of each event including date; areas affected; and total insured (industry) cost in “original” dollars. Unfortunately, no equivalent record of economic losses exists in Australia. Factors contributing to losses over time need be considered when estimating the current loss that would be sustained if each event in the Disaster List were to reoccur today. By way of example, imagine a repeat of Tropical Cyclone Tracy, which destroyed the city of Darwin in the Northern Territory. Along with inflation, the population of Darwin has increased considerably since Tracy made landfall in 1974, as has the number of dwellings. There has also been an increase in both the average dwelling price and average wealth per person. Building standard amendments specifying more wind resistant construction designs were also introduced following Tracy and as a result of this, the general construction quality in Darwin (assuming the code has been enforced) will be higher today than in

29 At present 45 Indian Rupees are equal to a U.S. Dollar but the exchange rate has varied considerably over the years. Hence conversion of the above figures to U.S. Dollars may be misleading. 30 Raghavan S. and A. K. Sen Sarma, 2000: Tropical cyclone impacts in India and neighbourhood. Storms, Vol. 1, R. A. Pielke Sr. and R. A. Pielke Jr., Eds., Routledge , 339–356.

539

1974. An indexation (normalization) methodology incorporating two surrogate factors to account for changes in population, inflation, and wealth since the date of the original event has been applied to event losses in the Disaster List (Crompton 2005)31. The approach is based on changes in both the number and nominal value of dwellings over time, where the nominal dwelling value excludes land value. These adjustments grow rapidly, for example, the nominal value of new dwellings in Australia increased by about a factor of 11 over the past 30 years. Indexed tropical cyclone losses were further adjusted, where appropriate, to account for the greatly improved building standards in tropical cyclone-prone areas that occurred in the early 1980’s (work in progress – upcoming publication by Ryan P. Crompton). As a result of the improved wind resistant design of housing, dramatic reductions in tropical cyclone-induced losses were observed following Tropical Cyclones Winifred (1986) and Aivu (1989) (Walker 1999)32 and more recently, Larry (2006) (Guy Carpenter 2006; Boughton et al. 2006)33,34. Figures 5.2.2a and 5.2.2b show annual losses aggregated by original and current (2006) loss values for weather-related events in the Disaster List. When indexed, the time series of insured losses exhibit no obvious trend over time. Figures 2a and b show that the increasing trend in unadjusted losses is largely attributable to changes in the: number of dwellings; nominal values of dwellings; and building standards in tropical cyclone-prone regions. Annual losses have been calculated for seasons ending 30 June to take account of the seasonality of the main meteorological hazards. The analysis begins from the 1966 season (1966/67) and ends at the current 2005 season.

31 Crompton, R.P. 2005. Indexing the Insurance Council of Australia Natural Disaster Event List. Report prepared for the Insurance Council of Australia, Risk Frontiers. 32 http://www.aon.com.au/pdf/reinsurance/Aon_Designing_Disasters.pdf 33http://gcportal.guycarp.com/portal/extranet/popup/pdf/GCPub/Tropical%20Cyclone%20Larry%20051906.pdf?vid=1 34 http://www.abcb.gov.au//documents/General/CTS_summary_TCLarry.pdf

540

0

200

400

600

800

1000

1200

1400

1600

1800

2000

1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005

Season

Orig

inal

Los

s ($

M)

Figure 5.2.2a: Original annual aggregate insured losses ($M) for weather-related events in the Disaster List for 12-month seasons ending 30 June.

541

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005

Season

Cur

rent

Los

s as

at 2

006

($M

)

Figure 5.2.2b: As for (a) above but losses have been indexed to 2006 dollars. Tropical cyclone losses have been adjusted for changes in building standards. While the indexation methodology does not incorporate every factor that may contribute to natural disaster losses, for example, how El Niño - Southern Oscillation (ENSO) affects tropical cyclone activity in the South-Western Pacific region, it does include the most significant ones. The importance of wind code adjustments in the Australian context should not be underestimated. For Tropical Cyclone Tracy, the indexed loss is halved after this is made, with a unique adjustment applied to each event. Analysing the tropical cyclone losses in the Disaster List alone tells us little about factors contributing to losses owing to the limited number of losses resulting from Australia’s sparsely populated tropical cyclone exposed coast. To explore tropical cyclones specifically, the focus is shifted to the time series of events that crossed the east coast of mainland Australia during the last 45 years. Only events having a central pressure less than or equal to 995hPa were included. The analysis begins from the 1961 season (1961/62) and ends at the current 2005 season. Figure 5.2.3 shows successive five-season period frequencies of tropical cyclones that have crossed the east coast. Within each of the five-season periods there are different numbers of El Niño, La Niña, and neutral events. With the exception of 1971-75, there have been between two and four tropical cyclones for each period. It comes as no surprise that La Niña episodes dominated during this period.

542

0

1

2

3

4

5

6

7

8

9

10

1961-65 1966-70 1971-75 1976-80 1981-85 1986-90 1991-95 1996-2000 2001-05

Five-season periods

Num

ber o

f eve

nts

Figure 5.2.3: Number of tropical cyclones to cross the east coast during five-year periods. Figure 5.2.3 only tells part of the story. Of more interest is the combination of frequency and intensity. According to Gray (2003)35, when normalized for coastal population, inflation, and wealth per capita, tropical cyclone-spawned damage in the U.S. rises by a factor of four for each successive increase in Saffir-Simpson intensity category. Thus a landfalling Category-3 hurricane typically causes about four times the normalized damage of a Category-2 hurricane and so forth. Assuming that this same ratio between cyclone categories holds true for Australian conditions, Figure 5.2.4 takes this weighting into account. In calculating these figures, Category-5 and -4 events have each been assigned an equal weighting of 1/4; Category-3 a weighting of 1/16; Category-2 a value of 1/64 and Category-1 events 1/256. These are then summed for each 5-year period to obtain a relative potential destructiveness index. By concentrating on the damage potential of the hazard alone, Figure 5.2.4 assumes a uniform portfolio of assets at risk. It also allows for Australia’s low population density and the large physical distances between population centres on the exposed east coast. Actual damage arising from individual tropical cyclones will vary widely as a result of differences in population, terrain, topography, proportions of construction conforming to improved (wind loading) building standards, wealth per capita, direction and forward speed of the tropical cyclone, storm surge, and rainfall. Again we see no obvious change (increase or decrease) in the potential destructiveness over the time periods represented in Figure 5.2.4. Figures 5.2.3 and 5.2.4 focus on the east coast because this is where most of the exposure is located; however, similar results hold for the western and north coasts of Australia or for the entire coastline. Nonetheless we do acknowledge that the small number of cyclones per five–year time interval makes it difficult to draw very robust conclusions.

35 Gray, W.M. 2003: Twentieth century challenges and milestones, In Hurricane! Coping with Disaster edited by Robert Simpson, American Geophysical Union, Washington DC 2003.

543

Figure 5.2.4: Combined frequency-severity of tropical cyclones that have crossed the east coast of Australia during five-year periods.

c) United States (prepared by R. Pielke, Jr., J. Gratz, and E. Faust) Consider economic damage (adjusted for inflation) related to hurricane landfalls in the United States, 1900–2005, as shown in Figure 5.2.5. Although damage is growing in both frequency and intensity, this trend does not reflect increased frequency or strength of hurricanes. In fact, while hurricane frequencies have varied a great deal over the past 100+ years, they have not increased in recent decades in parallel with increasing damages. To the contrary, although damage increased during the 1970s and 1980s, hurricane activity was considerably lower than in previous decades.

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18

0.2

1961-65 1966-70 1971-75 1976-80 1981-85 1986-90 1991-95 1996-2000

2001-05

Five-season periods

Freq

uenc

y-se

verit

y w

eigh

ting

544

Figure 5.2.5. Trend in U.S. hurricane damage, 1900-2005. Source: NOAA/NHC To explain the increase in damage, it is therefore necessary to consider factors other than variability or change in climate. Society has changed enormously during the past century and coastal development has taken pace at an incredible pace. Given the significance of societal change in trends of hurricane damage, one way to present a more accurate perspective on such trends is to consider how past storms would affect present society. We developed a methodology for ‘‘normalizing’’ past hurricane damage to present day values (using wealth, population, and inflation).

36 Figures 5.2.6a and 5.2.6b shows the historical losses of Figure 5.2.5

normalized to 2005 values. The normalized record shows that the impacts of Hurricane Andrew, at close to $53 billion (2005 values) (unpublished analysis by author, updated from Pielke and Landsea, 1998), would have been far surpassed by the Galveston, TX hurricane of 1900 which would have caused an estimated $159.8 billion damage or the Great Miami Hurricane of 1926, which would have caused an estimated $148.1 billion damage (preliminary estimate, work in progress) had it occurred in 2005, exceeding similarly accounted costs of Katrina. We can have some confidence that the normalized loss record accounts for societal changes because, unlike the unadjusted data, the adjusted damage data accurately reflect well-understood patterns of climate variability, such as the signal of El Niño and La Niña in hurricane frequencies.

37

36 Pielke, Jr., R. A. and C. W. Landsea, 1998. Normalized Hurricane Damages in the United States: 1925-95. Weather and Forecasting, American Meteorological Society, Vol. 13, 621-631. http://sciencepolicy.colorado.edu/admin/publication_files/resource-168-1998.11.pdf 37 Katz, R.W., 2002. Stochastic modeling of hurricane damage. Journal of Applied Meteorology, 41:754-762. http://www.isse.ucar.edu/HP_rick/pdf/damage.pdf

Total Losses per Year from Atlantic Tropical Cyclones in 2005Dollars

0

25

50

75

100

125

1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000

Bill

ions

Year

545

Figure 5.2.6b shows the historical losses from Munich Re’s database since 1954 normalized by the Pielke/Landsea methodology to 2005 values. In most of these years – with the exception of 1960, 1979 and 2005 – the differences are not large. In particular for hurricane Donna in 1960 Munich Re provides a historical loss estimate of US$1,250 m compared to US$397 m in the NHC data. Equally for 2005 Munich Re has some US$ 125 bn for economic losses of Katrina only instead of US$107 Billion in the NHC dataset, which accounts for the 2005 difference.

546

PL05 Normalized Losses per Year from Atlantic Tropical Cyclones(11-year centered average)

0

25

50

75

100

125

150

175

1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000

Bill

ions

Year

Figures 5.2.6a, 5.2.6b: The upper graphic (6A) displays estimated hurricane damages 1900-2005 if storms of the past made landfall with coastal development of 2005. A filtered curve is given by the black solid line. Source: Roger A. Pielke, Jr., work in progress. The bottom graphic (6B) displays the same calculated from Munich Re’s base loss data (1954 – 2005). The black solid line shows a 10-year running mean. Source: Munich Re 2006.

20

40

60

80

100

120

140

160

1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000year

loss

[US$

bn]

547

5.2.4: Differing views of the role of global warming on losses As indicated in the above case studies, in many locations around the world, especially in the Atlantic basin, economic losses from tropical cyclones has increased dramatically in recent decades even after accounting for the effects of inflation. This has led to some discussion of the relative role of changes in storms themselves in the context of rapid growth in wealth and population in locations exposed to tropical cyclone risks. While this subject will no doubt be covered in greater depth in other parts of the IWTC, we though it important to include a brief discussion of the subject as related to losses. Research on tropical cyclones has advanced rapidly since the last IPCC assessment. In 2005 Massachusetts Institute of Technology’s Kerry Emanuel published a study in the journal Nature that described an increase in the intensity of hurricanes in the North Atlantic and North Pacific.38 Another prominent study by Webster et al. has found an increase in the proportion of the strongest storms since 1970.39 These two papers, published in the midst of a record Atlantic hurricane season in terms of both activity and damages have prompted much discussion about trends in tropical cyclone behavior. Both the Emanuel and Webster et al. papers have prompted a vigorous discussion with responses to each subsequently published by Landsea40 and Chan41, respectively. A vigorous debate continues within the community about the trends themselves. Support for the findings of Emanuel and Webster et al. a study of globally integrated power dissipation of TCs based on reanalysis data.42 In contrast, a paper in Geophysical Research Letters in May, 2006 which find no trends in global tropical cyclone intensity from 1986-2005, with the exception of a dramatic increase in intense storms in the Atlantic Basin.43 Its author concludes,

These findings indicate that there has been very little trend in global tropical cyclone activity over the past twenty years, and therefore, that a large portion of the dramatic increasing trend found by Webster et al. [2005] and Emanuel [2005] is likely due to the diminished quality of the datasets before the middle 1980s. One would expect that if the results of Webster et al. and Emanuel were accurate reflections of what is going on in the climate system, than a similar trend would be found over the past twenty years, especially since SSTs have warmed considerably (about 0.2°C – 0.4°C) during this time period.44

Given these various results, a position paper by the World Meteorological Organization’s Commission on Atmospheric Sciences, its Tropical Meteorology Research Program Panel (whose authorship included Emanuel, Holland, Knutson, Landsea, among other prominent scientists) concluded:

38 Emanuel, K. 2005. Increasing destructiveness of tropical cyclones over the past 30 years. Nature 436:686–688. 39 Webster, P.J., G.J. Holland, J.A. Curry, and H.R. Chang. 2005. Changes in tropical cyclone number, duration, and intensity in a warming Environment. Science 309:1844–1846. 40 Landsea, C. W., 2005. Hurricanes and global warming, Nature, 438:E11-13. http://www.aoml.noaa.gov/hrd/Landsea/landseanaturepublished.pdf 41 Chan, J. C. L. 2006. Comment on "Changes in Tropical Cyclone Number, Duration, and Intensity in a Warming Environment," Science, 311:1713. http://www.sciencemag.org/cgi/content/full/311/5768/1713b 42 Sriver, R., Huber, M. 2006. Low Frequency Variability in Globally Integrated Tropical Cyclone Power Dissipation. GRL 43http://tropical.atmos.colostate.edu/Includes/Documents/Publications/klotzbach2006.pdf 44http://tropical.atmos.colostate.edu/Includes/Documents/Publications/klotzbach2006_talkingpoints.pdf

548

The research issues discussed here are in a fluid state and are the subject of much current investigation. Given time the problem of causes and attribution of the events of 2004-2005 will be discussed and argued in the refereed scientific literature. Prior to this happening it is not possible to make any authoritative comment.45

However, it seems reasonable to conclude that, at the present time, while detection of trends in tropical cyclone behavior may yet achieve a scientific consensus, attribution of such trends to human causes remains to be settled in the scientific literature. This area of science is undergoing rapid change as new research results are published, so it may be that detection and attribution will soon be unambiguously achieved. But until that occurs claims of definitive detection and attribution are premature, and thus so too would be any definitive link between trends in damage and human effects on tropical cyclones. Even with the unsettled state of the science, differences exist in how to interpret recent loss trends, even among the contributors to this report. The below subsections provide two different perspectives on the role that global warming may have on recent loss trends in the United States. Perspective #1: Little evidence for global warming effects on losses (prepared by R. Pielke, Jr.) It is tempting to look at aggregate loss data and conclude that higher August, September, October (ASO) SSTs cause increased U.S. damage. For instance, for the period 1950-200546 the top 19 warmest SST years have an average of $12.5B in damage, the next 19 have an average of $9.0 billion, and the next 18 have an average of $6.3 billion. But this pattern depends entirely upon the influence of the large losses of 2005. Through 2004 the results of such a binning are $7.2B, 9.0B, and 6.3B respectively. The nature of loss data, and its highly skewed distribution in particular, makes analyses with untransformed data problematic when seeking to establish robust statistical relationships. The Spearman (rank) correlation coefficient between ASO SSTs and damage 1950-2005 is low at 0.098. Figures 5.2.7a and 5.2.7b shows the lack of meaningful relationship between normalized U.S. hurricane damages (NHC data, transformed with the natural log) and North Atlantic sea surface temperatures 1950-2005 and 1950-2004.47 The r-squared values are low with or without 2005 included, and the regression results are not statistically significant (p = 0.28 and 0.69 respectively). There is consequently no systematic evidence that higher SSTs are systematically associated with larger losses. One reason for the lack of a significant relationship is that any signal that exists between SSTs and hurricane behavior is lost in the randomness of each relationship in the sequence of: ∆Basin SSTs ∆Basin PDI ∆Landfall PDI ∆Economic damage48 So while there is undoubtedly a relationship between SSTs and hurricane activity, it may nonetheless be difficult to observe any relationship between trends in SST, trends in hurricane behavior, and trends

45 http://www.bom.gov.au/info/CAS-statement.pdf 46 We use this period because it avoids many years pre-1950 in which there were no recorded damages. This is due to the lack of development on the coast, rather than a lack of hurricanes. Such non-damage years skew the analysis. If one looks only at years in which damage occurred then the relationships for 1950-2005 hold in the earlier period. 47 http://www.cpc.noaa.gov/data/indices/sstoi.atl.indices

48 Thanks to Kerry Emanuel for proposing this explanation.

549

in economic damage data. In the loss vs. SST data, 2005 clearly stands out as an outlier from the rest of the dataset. Expectations that future hurricane seasons will look more like 2005 rather than the rest of the dataset may turn out to be correct, but such expectations are not supported by the historical record of the relationship of SSTs and damage. Consequently, it is premature to attribute any part of the historical trend in losses to global warming, though such a connection may be made in the future. As the Hohenkammer Workshop concluded:

Because of issues related to data quality, the stochastic nature of extreme event impacts, length of time series, and various societal factors present in the disaster loss record, it is still not possible to determine the portion of the increase in damages that might be attributed to climate change due to GHG emissions.49

49http://sciencepolicy.colorado.edu/sparc/research/projects/extreme_events/munic

h_workshop/workshop_report.pdf

550

Figure 5.2.7a and 5.2.7b. Damage versus August, September, October (ASO) North Atlantic SSTs. 1950-2005 (a) and 1950-2004 (b), R. Pielke, Jr.

(a)

(b)

551

Perspective #2: Remarkable evidence for global warming effects on losses (prepared by E. Faust) The running means in Figures 5.2.6a and 5.2.6b show no obvious longitudinal trend in losses related to tropical cyclones, even with the extreme losses of 2004 and 2005. But if analyzed more closely, the normalized loss data show nonetheless systematic changes over time. Fundamental to these changes is the presence of a correlation between normalized annual losses and June-October annual tropical sea surface temperatures. Munich Re analyzed the respective annual SST anomalies and annual normalized losses since 1900. Figure 5.2.8 simply displays the normalized losses against the SST anomalies. Also, the average loss calculated for a running window of 0.2°C in width is displayed (red line). The running average is shown over a range where the 0.2°C windows are populated densely enough (at least 12 data points, i.e. half the maximum population, see the dashed black line). A remarkable general increase in average annual normalized losses with increasing SST can be observed over the -0.4°C to +0.4°C anomaly range. Spearman’s rank correlation coefficient, which is independent of the distributions involved, gives 0.26 for the range from -0.4°C to +0.4°C and 0.28 for all of the data. This positive correlation prompted further analysis. In order to perform a regression analysis on the data we need to transform the loss data, which are heavily skewed (skewness of 4.03), into a Gaussian distribution. This was done by applying the natural logarithm and cutting the data at the smallest positive loss – thereby excluding all years with zero losses from the distribution (including them would have retained the strong skewness). The result of the regression analysis is displayed in Figure 5.2.9, which depicts an increase in losses with SST – with a strong scatter of data, as has to be expected, and therefore only a small R-square. The slope of the regression line is approx. retained if the high LN-value for 2005 is excluded from the data (the slope changes from +0.14 to +0.11) – but this need not be done because the data conform to a Gaussian-like distribution. The scatter is accounted for by a strong year-to-year variability in the ratio of landfalling storms to basin-wide storms, by varying landfall locations and loss amounts, and by effects of natural climate variability like ENSO: El Nino episodes would weaken hurricane activity regardless of SSTs in the Atlantic, La Nina episodes would have the opposite effect. It should be remembered that years with zero losses are excluded from the data in Figure 5.2.9 in order to achieve an un-skewed distribution. Hence, considering the fact that the percentage of years with zero losses in the cold half of the SST spectrum (21%) is more than double the percentage in the warm half (9.1%) (13 zero years in the cold half of the SSTA range from -0.68°C to +0.11°C in 62 years versus only 4 zero years in the warm half from +0.11°C to +0.91°C in 44 years), it turns out that the increase in losses with SST is even larger – as depicted by the reduced data in the scatter plot of Figure 5.2.9. Taken together with arguments and observations from meteorology –warmer oceans foster more intense and, in the North Atlantic, more frequent storms in the long term – the possibility cannot be ruled out that tropical SSTs act as an additional driver for increasing losses besides changing societal exposures and vulnerabilities. .

552

Figure 5.2.8 Mean annual normalized US hurricane losses per SST anomaly (blue points, projecting on left-hand axis) (Hadley Centre data set for 10°N–20°N, east of 80°W). Red curve indicates the annual loss average calculated over a running window of 0.2°C in width (projecting on left-hand axis); dashed black line displays the count of data points per running 0.2°C window (projecting on right-hand axis). For a more detailed explanation, see the text. Source: Munich Re 2006, work in progress.

553

Figure 5.2.9 Natural logarithm of annual normalized US hurricane losses since 1900 versus SST anomalies of the tropical North Atlantic (Hadley Centre data set for 10°N–20°N, east of 80°W). Years without any losses were omitted in order to obtain an un-skewed Gaussian distribution by logarithmic transformation, which is a precondition for performing a linear regression analysis. Source: Munich Re 2006, work in progress. Even using the much shorter CPC SST time series, which covers only the years from 1950 onward, results in almost the same increasing slope of the regression line. Transforming the normalized losses by applying the natural logarithm in the original data yeilds a Gaussian-like distributed data and a positive slope of the regression line (Figure 5.2.10). We get almost the same slope (0.13 vs. 0.14) and the same small order of R-square as with the SST data since 1900. So making use of the shorter CPC SST time series (from 1950 onward) produces results which are quite consistent with those using the longer time series of SSTs from the Hadley Centre.

y = 0.1417x + 21.384R2 = 0.0368

10

15

20

25

30

-8 -6 -4 -2 0 2 4 6 8 10SSTA [in units 0.1蚓 ]

LN(a

nnua

l nor

mal

ized

US

TC L

oss)

SSTA [in units 0.1 oC]

554

warm phase (n = 56 years)mean median std devUS$ 13.1 bn US$ 3.9 bn US$ 25.7 bn

cold phase (n = 47 years)mean median std devUS$ 5.1 bn US$ 0.5 bn US$ 12.2 bn U-test acc. to WILCOXON/MANN/WHITNEY: both loss-frequency distributions and respective median values are different in a statistically significant way (α = 1%). (ζ = 2,93 > zα=1% = 2,33) Table Properties of loss-frequency distributions of annual losses from warm and cold phase years. Data base: Pielke/Landsea normalized US TC losses. Source: Munich Re 2006, work in progress.

cold phase years

warm phase years

> US$ 1 bn19 (of 47) 40%

36 (of 56) 64%

> US$ 5 bn10 (of 47) 21%

26 (of 56) 46%

> US$ 10 bn8 (of 47) 17%

17 (of 56) 30%

Table Percentages of years exceeding specified annual loss thresholds in warm and cold phases of the 20th century. Data base: Pielke/Landsea normalized US TC losses. Source: Munich Re 2006, work in progress.

Figure 5.2.10 Natural logarithm of annual normalized US hurricane losses since 1950 versus SST anomalies of the North Atlantic (CPC ASO SSTs). The year 1958 without any loss is omitted in order to obtain an un-skewed approx. Gaussian distribution by logarithmic transformation, which is a precondition for performing a linear regression analysis. Source: Munich Re 2006, work in progress.

y = 0.1319x - 15.186R2 = 0.0311

10

15

20

25

30

268 270 272 274 276 278 280 282 284 286SST [units 0.1蚓]

LN(a

nnua

l nor

mal

ized

US

TC lo

ss)

SST [units 0.1 oC]

555

By analyzing the SSTs of the tropical North Atlantic phases of – on average – warmer (1926–1970; 1995–today) and cooler water temperatures (1903–1925; 1971–1994) can be identified. Given the demonstrated correlation of mean annual US losses and tropical cyclones on SSTs in the tropical North Atlantic, these SST phases should project on phases of different loss frequency distributions over the respective phases, which in turn materialize in differing distribution means. In order to check this hypothesis, a comparison of the frequency distributions of normalized annual TC losses in the warm phase years and the cold phase years of the 20th century was performed. The analysis assumes loss data covering all of the 20th century, so that normalized NHC data were chosen. As a result, median and mean values are much higher in the warm phase distribution, and these distributions and the respective medians are different in a statistically significant way (α = 1 %; the WILCOXON-MANN-WHITNEY test was chosen which has no requirements regarding data distribution). In addition one can find that the percentage of years exceeding specified loss thresholds – US$ 1bn, 5bn and 10bn – is much higher in warm phases than in cold phases. So even if there is no year-to-year increase in loss amount in the normalized data, there is a shift in terms of the loss distributions accompanying periods of cooler and warmer tropical waters. As seen before the percentage of years with zero losses in the cold half of the SST spectrum (21%) is more than double the percentage in the warm half (9.1%) – it seems more likely to have years with no losses while SSTs are cooler than on average. We found a general increase in mean annual normalized TC losses with increasing SSTs as visualized by the running mean in Figure 5.2.8 (rank correlation coefficient of 0.3) and the linear regression analysis identified positive correlations of SSTAs (Hadley Centre and CPC data) and logarithmically transformed US normalized TC losses. Admittedly the data scatter is quite substantial resulting in small R-square, but this has to be expected. Taken together with climatological evidence on changing intensities/frequencies of TCs with changing SST in the tropical North Atlantic, our findings are in line with conclusion no. 10 of the Hohenkammer Workshop:

There is evidence that changing patterns of extreme events are drivers for recent increases in global losses.50

Hence, the conclusion cannot be ruled out that if the increases in tropical Atlantic SSTs were to continue in the long term due to anthropogenic climate change51, we would have to expect a shift towards hurricane loss distributions with ever increasing high-loss portions. This would be an additional case of increase besides the strong impact of increasing exposures and vulnerabilities on increasing losses due to societal changes over time.

50http://sciencepolicy.colorado.edu/sparc/research/projects/extreme_events/munich_workshop/workshop_report.pdf 51 Barnett, T. P. et al. (2005), A Warning from Warmer Oceans, Science 309, S. 284–287

556

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES Topic 5.3 : Mitigation Strategies and Community Capacity Building for Disaster Reduction Rapporteur: Linda Anderson-Berry Bureau of Meteorology, Australia Email : l.anderson-berry@bom.gov.au Working Group: David King, Hilda Lam, Terry Hart, Mike Bergin 5.3.1 Introduction Tropical cyclones, like all other naturally occurring hazard events have the potential to severely affect the lives and livelihood activities of people in the communities they impact. Human suffering and social, economic and environmental loss continue to be an almost inevitable outcome of land-falling tropical cyclones. Alarmingly over time, and despite ever-improving technological solutions for forecasting, detecting, identifying the physical dimensions and monitoring development and movement of tropical storms, and communicating warnings messages, this loss continues to increase. Additionally, in the face of changing global climate regimes and the likelihood of more frequent, and possibly more intense, tropical cyclones, coupled with growing populations in tropical coastal regions, it is likely that more people in the tropical regions will be in harms way and levels of loss and suffering will continue to escalate. To mitigate such loss, warnings systems have been developed throughout the WMO community in the context of a total warning system. This extends the concept of warnings from the simple delivery of a message about impending severe weather conditions to a process that begins with using the best science available to predict and monitor the development and progress of, for example, a tropical storm, to the production and delivery of a timely and accurate message to a receptive, prepared and resourced community in a format that is understood. Thus, empowering society – at all levels of social aggregation, from central governments down to individuals in households to make informed risk minimizing decisions and that appropriate loss protective and defensive actions. Simply stated this means that the warning can only be considered to have been successfully delivered when the community on which it is focused has had the capacity to respond. Understanding and supporting the building of community capacity has evolved as a dominant theme in disaster mitigation. Strategies and research emphases for understanding natural hazards, such as tropical cyclones, and their effect on human communities with an ultimate view to mitigating loss has seen a paradigm change over time. Until the early 1990’s emphasis (and research effort) was focused on understanding the hazard itself – an essentially and predominantly physical science approach. This emphasis began to change throughout the 1990’s - the International Decade for Disaster Reduction – IDNDR when the focus shifted towards attempting to understand the impacts. Research and practice to this end saw an inclusion of research from within the social sciences and an investigation of mitigation and adaptation strategies. Identifying, understanding and reducing community vulnerabilities was seen to be important and disaster mitigation solutions began to be based on the development of social policy as well as engineering defences. This direction was reflected in presentations at IWTC-V. Following on from the IDNDR has been the United Nations initiated International Strategy for Disaster Reduction (ISDR). This strategy has overseen the development and implementation of the Hyogo Framework for Action 2005-2015 – Building the Resilience of Nations and Committees to Disasters and

557

the Platform for the Promotion of Early Warnings (PPEW). Community capacity building has become a dominant theme of ISDR, particularly in relation to disaster risk reduction, developing strategies to mitigate loss of life and property in the event of natural disasters and developing early warning systems. The goal is to support a ‘bottom-up’ approach to strengthening and building resilient communities with the capacity to prepare for, mitigate, respond to and recover from natural hazard events. In the case of hazards of meteorological hydrological and oceanographic origin this has supported the building of strong partnerships between the weather services that have primary responsibility for developing and delivering warnings to the full range of stakeholders in government, emergency services, educational services, community management and citizens (individually and collectively) In this summary paper – disaster mitigation strategies in the context of community capacity building will be discussed – with the support of case study examples Capacity building Capacity building – both as a term and a principal is widely used but perhaps not so widely understood, and is rarely clearly defined. The United Nations Development Programme (UNDP) defines Capacity – in the development context as:

• the ability of individuals, organisations and societies to perform functions, solve problems and set and achieve goals. And

• its development entails the sustainable creation, utilization and retention of that capacity in order to reduce poverty, enhance self reliance and improve peoples lives

For the purposes of this paper community capacity building in support of disaster mitigation is defined as “……. strategies that enhance or build the ability of the human population to make effective risk minimizing decisions and take effective risk minimizing actions that will increase the levels of safety. This includes building human capacity to interact with and utilize the technology as well as supporting the development of strongly networked cohesive communities.” NMHS’s generally support community capacity building and serve their ‘user’ communities as effectively as is possible through the very strong partnerships and close and interlinked working relationships they have developed with their various stakeholders among government agencies, community service providers and particularly the emergency management community. 5.3.2 The Australian experience - Partnerships with Government and non-Government

Agencies

Queensland Tropical Cyclone Coordinating Committee

The Queensland Tropical Cyclone Coordination Committee (QTCCC) was established in 1995 to provide advice to the State Government’s Disaster Mitigation Committee on measures to mitigate the effects of tropical cyclones on Queensland communities. The QTCCC’s role encompasses disaster mitigation research, policy development, disaster risk assessment (hazards and vulnerabilities of communities), mitigation measures, community preparedness and post disaster assessment in relation to tropical cyclones.

The QTCCC’s functions are to: enhance community safety, mitigation and prevention capability across all Queensland communities in relation to the impact of tropical cyclones; contribute to the National, State and local strategic policy framework relating to the impact of tropical cyclones; advise on roles, responsibilities and priorities of entities involved in the management of tropical cyclone impacts; and review tropical cyclone impacts to assess mitigation processes.

The Committee is jointly chaired by the Director of the State Government’s, Disaster Mitigation Unit in

558

the Department of Emergency Services (DES) and the Regional Director, Australian Bureau of Meteorology (BoM). Committee membership includes representatives from State Government Departments of Health, Transport and Local Government and Planning, Sport and Recreation; the Environmental Protection Agency; Queensland; Queensland Police Service; the Institute of Public Works Engineers Australia, Queensland Division; the Planning Institute of Australia; Tourism Queensland; Insurance Disaster Response Organisation; Queensland Local Government Association; Emergency Management Australia (Australian Government); and researchers from Geoscience Australia and James Cook University’s Centre for Disaster Studies and Cyclone Testing Station other specialist advisers are invited as required. The Committee develops and maintains a three-year rolling strategic plan and annual work plan for each year. One of it’s major activities is the coordination of the annual pre-season public awareness and preparedness education campaign when Collectively and cooperatively just prior to the beginning of the Tropical Cyclone the Bureau and its partner agencies present a series of interactive demonstrations, workshops and meetings during an annual ‘road-show’. The Bureau’s Queensland Regional Director, senior forecasters, hydrologists, together with key executive staff from the Emergency Management Queensland’ State and regional offices and other partner agencies travel to major townships and cities in tropical cyclone prone northern Queensland to provide information about seasonal forecast, tropical cyclone warning products, and a holistic public education campaign to local and regional emergency services, local government groups, schools, community services groups and the general public. This annual direct interaction with residents in cyclone-prone communities has proved to be a powerful tool in raising community awareness and building community trust and confidence in the weather and emergency services. 5.3.3 The Hong Kong experience - Coordination with emergency services and public education activities To strengthen the community’s capability to withstand tropical cyclone (TC) hazards, the Hong Kong Observatory (HKO) recognizes the importance of good coordination with the emergency response units as well as a long-term programme for public education and outreach. To ensure good coordination between HKO with other government departments and non-government agencies so as to trigger fast response actions in TC situations, a contingency plan for natural disasters is in place in Hong Kong stipulating the triggering mechanism of the warning system, the responsibilities of various agencies in responding to natural disasters. The HKO also holds annual liaison meetings with key departments such as the education, transport, drainage and geotechnical control authorities to review arrangements for emergency response. Government-wide seminars are also conducted every year to enhance understanding and cooperation of government emergency response staff. The HKO organizes public education and outreach activities such as public talks and lectures, school visits, exhibitions, TV documentary series and publicity campaigns to raise public awareness of disaster risks. Through these activities, the following information is provided to facilitate better understanding of the TC phenomena and warning service: meaning of the TC warnings; issuance and cancellation criteria; advisories and precautionary measures; rationale behind criteria and recommended actions; explanations of the nature of TC and its impacts; descriptions of cases of extreme events, both local and worldwide; information about the effectiveness of precautionary actions; challenges in tropical cyclone forecasting. Public education and outreach activities are scheduled throughout the year: ♦ In the quiet season, visits to the observatory by students, open day, exhibitions in public venues and

outreach visits by Observatory staff to homes for the elderly and schools, campaigns to raise awareness, including various related contests involving particularly students are organized.

559

♦ Prior to the start of the tropical cyclone season, the Director of the Observatory appeals to the public to be prepared for natural hazards. Regular announcements on TC through TV and radio are broadcast. The HKO also meets with stakeholders to refresh their memories and advise them about service enhancements.

♦ During the tropical cyclone season, broadcasts of announcements through TV and radio are stepped up. The HKO meets with those users hardest hit to understand their problems to identify quick fixes and long-term solutions.

♦ After the tropical cyclone season, HKO meets with users to discuss enhancements to services, organize training courses for the public and special user groups.

In 2004, HKO organized a Symposium on Natural Disaster Planning and Preparedness which provided an opportunity for meteorologists, disaster managers and academics in the region to exchange experience and strengthen co-operation in disaster preparedness and mitigation. In 2005, the HKO and other government departments as well as NGOs jointly organized a year-long public education campaign, called “Safer Living – Reducing Natural Disasters”, to promote public awareness and understanding of natural hazards. The year-long campaign includes a four-episode TV programme on safer living, a meteorological series TV programme, a tropical cyclone naming contest, popular science lectures, bookmark design contests, a seminar on natural disaster reduction and a large exhibition with rescue drill demonstrations. The “TC Name Nomination Contest” which invited the public to nominate names for tropical cyclones in the northwest Pacific and the South China Sea attracted more than 20,000 entries. HKO is also keen in identifying collaboration partners for public education such as the Hong Kong Education City for which the target audience are students and teachers. Web pages for educational resources and publicity materials on TC are developed to facilitate the spread of the knowledge in the education sector. Partnership with the media contributes to the effective promotion of public education on the hazards of TCs to a wide range of audience. Through a sustained programme of education and outreach, the public and all emergency response units involved can gain a better understanding of the characteristics of the hazardous phenomena, where vulnerability lies, and what their respective actions should be on receiving warnings, thereby enhancing the resilience of the community to TC hazards. 5.3.4 The Fiji Experience - Capacity Building in the Pacific – supporting infrastructure and

training Over the past few years RSMC Nadi has suffered from staff shortages that has resulted in a staff shortages during the very active cyclone season and staff not able to be released for training. The Bureau of Meteorology Australia has been able to assist with attachments of staff for periods of 12 months or more – supported largely by Australia’s aid agency AusAID through it’s Pacific support projects aimed at providing support through infrastructure and training to build local capacity. Australian tropical cyclone and severe weather forecasters from the RSMC Darwin have provided forecasting support in RSMC Nadi. They have also assisted with training in Fiji and the Pacific Island nations of Vanuatu, Tonga, and Samoa. Infrastructure support has been provided with the implementation of Tropical Cyclone module that is used in forecasting operations in Australia. The TC Module is able to ingest NWP tracks and produce a consensus forecast which then can be modified. Graphical warning products that indicate past and forecast tracks and threat areas can then be generated. These have significant advantages over text products in conveying information to the public. Australia has successfully sought additional funding through AusAID to build on this capacity building initiative and extend support for the region, including the development of the TC Module, in the broader framework of Disaster mitigation planning recently adopted by the leaders of the Pacific Island countries WMO is developing severe weather projects to demonstrate an end-to-end approach from the NWP

560

and other numerical guidance products available from global and regional centers through the national meteorological services to end users such as emergency service organizations. One of these projects is specifically related to tropical cyclones. The concept is to explore both current and new types of NWP products that could assist end users particularly in extending the outlook period of their planning (eg through medium-range ensemble products that can give advance warning) WMO Region V through the Chair of its TC Committee expressed interest in participating in such a project. The initial phase, however, will be conducted for the south-east Asia region with the Pacific later. 5.3.5 Ranet Ranet (radio internet) is a project concerned with communication of TC (and other hazard) education and warning information to the local community level. The primary aim has been to build community capacity by ensuring information is delivered to ‘the-last-mile’. This was initiated by NOAA Office of Global Programs in Africa and has been extended to South east Asia and the Pacific with contributions from the US, UK, New Zealand and Australia. The implementation has explored affordable and sustainable communications options and is focusing on:

• Use of local FM radio stations that can broadcast to local communities • Digital HF to allow connection from a central site to communicate with remote communities

within a country and also for long distance international communication (eg for the transmission of observations onto the GTS where this has not previously been possible), and

• The WorldSpace digital sattelite broadcasts which is providing a oneway data broadband capability for countries able to receive its transmission; the broadcast currently includes web pages and meteorological data including the EMWIN content and warnings. There is scope to include other types of community education material such as health and agricultural advice and warnings.

Australia’s aid agency continues to support this highly successful capacity building project. Future developments will link with and be boosted by the telecommunications technology that is being implemented for the tsunami warning system (both data collection and transmission of warnings) for the Pacific region. In October 2005 at the South Pacific Forum the leaders of the Pacific Island Countries endorsed the Pacific Regional Disaster Risk Reduction and Disaster Management Framework 2006-2015 as part of the Pacific Plan. This represent the highest possible endorsement within the Pacific region of the disaster mitigation approach (implicit in this is capacity building) that underpins ISDR. The forum requested that regional policies and plans be implemented by the end of 2008 and asked for progress reports in 2006 and 2007. As meteorological, hydrological and oceanographic hazards (particularly tropical cyclones are the major component of natural hazards it is expected that NMHS’s will have a significant role to play in the implementation of the framework. The South Pacific Applied Geoscience Commission (SOPAC) is the regional representative of the ISDR – to which the Bureau and WMO are strongly committed. 5.3.6 Understanding and evaluating community capacity – the value of Post-impact data

collection and evaluation a) Tropical Cyclone Zoe Solomon Islands December 26-29, 2002 One of the outcomes of the WMO IWTC V, Cairns December 4-13, 2002, was a raised awareness of the need for scientists from all disciplines, specialising in tropical cyclones, to cooperatively and collectively evaluate the effectiveness of tropical cyclone warning systems and the impact of land-falling severe tropical cyclones on coastal communities. It was agreed that this should ideally be achieved through the conduct of (internationally) cooperative, multi-disciplinary post-impact case

561

studies. Within weeks Severe Category 5 Tropical Cyclone Zoe impacted the tiny Solomon Islands communities of Tikopia and Anuta. Between December 27 and 29, 2002 these tiny remote islands with a combined population of less than 1800 village dwellers and essentially subsistence economies and virtually ‘modern’ infrastructure were subjected to sustained hurricane force winds for periods in excess of 30 hours. Warnings were issued through the Solomon Islands Weather Service with the support of the Australian Bureau of Meteorology. However, the islanders had been without 2-way radio communication for several weeks and their ability to receive warnings via short wave radio was unknown. The force of the storm was relentless, dwellings and food crops were for the most part destroyed, canoes for fishing and near island travel were lost and the environment was devastated. National and international concern for the welfare of the islanders was high but due primarily to the absolute remoteness of the islands, it was almost 2 weeks before direct contact could be made, the fate of the people known. As part of the first phase of assistance provided to the Tikopian and Anutan communities a multi-agency team (that included representatives of the Solomon Islands Weather Service and the Bureau of Meteorology Australia) carried out an holistic post impact assessment. Miraculously, no-one had died and there were no serious injuries but the people were mostly without food, housing and clothing. It seemed that the ability to sustain life on the islands into the future was limited at best. However, village communities typically enjoyed strong cohesive family and societal networks that were based on a long history of adherence to customary practices and belief systems, they were also strongly networked with communities of Tikopians and Anutan’s that were currently living away from their home islands. Levels of ‘social capital’ in these communities were high and with support coming from within and from outside, these resilient communities have demonstrated a remarkable capacity to recover and slowly rebuild from what seemed to be an impossible situation. A full discussion of this study will be presented at IWTC-VI

Fig 5.3.1 FTikopia – Ravenga Village area

Landslides Vegetation stripped

Village washed away

Swamp taro destroyed

Fresh water lake in-filled and salt infiltrated

562

b) Tropical Cyclone Larry March 20 2006 Northern Australia Tropical Cyclone Larry was a Category 4 Tropical Cyclone that impacted the economically significant Innisfail region in Far Northern Queensland. It was the most destructive land-falling tropical cyclone to impact a populated Australian coastal community in more than 30 years. Several post-impact assessments have been carried out to measure impact for response and recovery support purposes, to measure the environmental and societal impact and to evaluate warnings effectiveness. Miraculously, despite widespread devastation, there were no deaths and only minor injuries. Post impact assessments discovered evidence of sound preparation amongst strong, well networked rural and small town communities. People sheltered in their own homes, of which approximately 20% were totally destroyed. The population in the storm surge zone was advised to evacuate prior to the storm and with levels of community trust in the weather and emergency services generally high, most people did as they were advised, finding shelter with friends and relatives, mostly in the same area, but back inland from the storm surge zone. Post disaster surveys showed that most of the adult population had experienced previous cyclones and were generally aware of the tropical cyclone risk. The communities were cohesive and strongly networked. People behaved sensibly, did the right things and sheltered inside their houses at advised to do so. It was an example of a prepared and experienced community with high levels of social capital and the capacity to fare very well. A more complete review of Tropical Cyclone Larry impact will be presented at IWTC-VI. 5.3.7 Summary ‘Community capacity’ is increasingly being recognised as a reliable indicator of how human populations are likely to respond to and recover from the impact of hazard events. However, understanding and measuring community capacity presents a challenge. Weather Services, community and emergency managers, academics and hazard research centres are all increasingly investing in research and activities aimed at investigating community capacity, and ultimately applying strategies to build community capacity in support of disaster mitigation. This summary paper has provided an overview discussion of capacity building, supported with examples provided by various case studies. It has considered capacity building in the context of supporting high-level Government and multi-agency policy forums such as QTCCC, and public education strategies such as those delivered through HKO. It has considered community capacity building in terms of infrastructure support and training with examples from the Pacific region and Ranet. Discussion has also focussed on identifying situations, in both a developed and developing world context, where high levels of social capital and strong community capacity has proven to support community resilience and disaster mitigation in the broadest sense. A fuller discussion will be provided through presentations at IWTC-VI. Land-falling tropical cyclones can destroy lives and livelihoods, devastate the environment, and crush local and even national economies. Impacts can be somewhat mitigated, and NMHS’s are dedicated to achieving the highest levels of disaster mitigation possible through the delivery of effective tropical cyclone warnings in the holistic context of a total warning system. To achieve this NMHS’s work in partnership with their emergency management stakeholders to build and support communities with the capacity to take actions that will minimise loss of life and suffering. Bibliography Anderson-Berry L. and King D. 2005. "Mitigation of the Impact of Tropical Cyclones in Northern Australia through Community Capacity Enhancement". Special issue of Mitigation and Adaptation Strategies for Global Change (2005) 10: 367–392 ed. E.Haque. Anderson-Berry L, King D & Crane G. 2002. Assessment of the Effectiveness of Various Methods of Delivery of Public Awareness Information on Tropical Cyclones to Queensland Coastal Communities.

563

Report on Project 05/2001 for Emergency Management Australia.

Davidson, J. and M.C. Wong, 2005: Guidelines on Integrating Severe Warning into Risk Management (PWS-13), WMO/TD No. 1292. Hoogenraad, Wouter, van Eden Ronald & King David. 2004. “Cyclone Awareness Amongst Backpackers in Northern Australia.” AJEM. Vol 19 No 2. King D. 2006. (In press: accepted 11/05)“Organisations in Disasters.” Special issue of Natural Hazards edited by R. K. Chadha King D. 2006. (In press). “Planning for Resilience”. Chapter in Paton, D. & Johnston, D. eds. Disaster Resilience: an integrated approach. Pub. Charles C. Thomas, Springfield, Illinois King D. and Gurtner Y. 2005. “After the Wave: A Wake Up Warning for Australian Coastal Locations”. Australian Journal of Emergency Management, Vol. 20 No. 1. King D., Cottrell A., Goudie D. and Cunliffe S. 2005. “Community Hazard Awareness and Resilience in Northern Australia”. In Know Risk: UN/ISDR. Official publication of UN WCDR, Kobe, January 2005. Published by Tudor Rose, Leicester, UK King D. 2004. “Understanding the Message: Social and Cultural Constraints to Interpreting Weather Generated Natural Hazards”. International Journal of Mass Emergencies and Disasters. Vol 22 No 1 pp 57-74 King D. 2006. Post Cyclone Monica Survey. Centre for Disaster Studies & Bureau of Meteorology King D. & Goudie D. 2006. Cyclone Larry Post Disaster Survey. Centre for Disaster Studies & Bureau of Meteorology King D., Cottrell A., Anderson-Berry L., Cunliffe S., MacGregor C., McLachlan E and Antrobus J. 22000011.. Cyclone and Natural Hazard Vulnerability in Remote and Indigenous Communities of North Queensland: Final Reports. CCeennttrree ffoorr DDiissaasstteerr SSttuuddiieess,, JJaammeess CCooookk UUnniivveerrssiittyy.. EEmmeerrggeennccyy MMaannaaggeemmeenntt AAuussttrraalliiaa Lam, C.Y., 2005: Disaster Risk Management – the Weather Perspective, presented at the Asian Conference on Disaster Reduction, Beijing, China, 27-29 September 2005. Rogers, D., 2005: Turning Crisis Management into Risk Management – The Role of Weather Forecasting, presented at the Seminar on “Safer Living – Reducing Natural Disasters”, Hong Kong, China, 17 October 2005. http://www.unisdr.org/wcdr/public-forum/reports/WS007.pdf#search=%22CAPACITY%20BUILDING%22 http://www.unisdr.org/wcdr/thematic-sessions/presentations/session5-3/ocha.pdf#search=%22CAPACITY%20BUILDING%22 http://severe.worldweather.wmo.int/ http://www.sere.ucar.edu/ http://www.unisdr.org/eng/hfa/hfa.htm http://www.ccb.ucar.edu http://www.tesag.jcu.edu.au/CDS/index.htm

564

WMO/CAS/WWW

SIXTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES Topic 5a : Report to the IWTC-VI On the PROGRAM FOR IMPROVEMENTS TO HURRICANE INTENSITY FORECASTS AND IMPACTS PROJECTIONS (HiFi) Rapporteurs: Greg Holland1 and Roger Lukas2 1National Center for Atmospheric Research, Boulder, Colorado, 2University of Hawaii–Manoa, Honolulu, Hawaii. Contributors: Eric Chassignet (Florida State University), Ahuyi Chen (IRSMAS), Cort Cooper (Chevron). Kerry Emanuel (MIT), Chris Fairall (NOAA). Isaac Ginis (University of Rhode Island), Mark Luther (Florida Southern University), Frank Marks (NOAA AOML), Tom Sanford (UW), Nick Shay (RSMAS), Alexander Soloviev (Nova Southeastern University) 092006 Executive Summary The Program for Improvements to Hurricane Intensity Forecasts and Impacts Projections (HiFi) will provide the basic and applied research needed to reduce the error in 48-hour intensity forecasts for hurricane-strength storms by at least 10 kt (approximately one half of a Saffir-Simpson category) within the next five years, with an emphasis on improved forecasting of rapid intensification and decay, and decay and reintensification cycles In addition Improved projections of likely hurricane activity over the next 50-100 years will be developed. An appropriately balanced program of observations, modeling and theoretical research is to be developed and the research results will be transitioned to operational status for short-term predictions and for hurricane impacts projections on longer time scales. The National Center for Atmospheric Research will be the lead organization for the HiFi program, working with university, government and industry scientists. A Science Steering Committee is in the phase of developing the Research and Implementation Strategy for HiFi. Implementation of the basic research agenda will take place during the first five years, and the transition to operational applications will be completed during the subsequent five years.

5.a.1 Background The devastating 2005 hurricane season in the Gulf of Mexico and the previous record four hurricanes striking Florida during 2004 have strongly underscored the need for improved prediction of hurricane risk and assessment of the longer-term risk level. The marked increase in impacts over the past decade have arisen from an increase in the frequency and intensity of hurricanes affecting US coastal and offshore resources combined with increasing populations, infrastructure and economic activities in vulnerable regions. This is already a major issue for the United States and projected future increases in vulnerability combined with potential increases in hurricanes require urgent action to minimize the impacts. Minimizing such hurricane impacts is a complex activity that requires attention to potential wind damage, heavy rain and flooding, wave and storm surge inundation, coastal erosion, and pollutant pathways, all of which depend critically on accurate forecasts of hurricane intensity and structure. The time scales range from forecasts over several days for immediate responses, to projections over years and decades to support proper planning and engineering design decisions.

565

A major commitment of resources for research and development in the 1990s initiated by the Office of Naval Research led to remarkable improvements in track forecasting skill, with the 48 h forecasts now better than those at 24 hours a decade ago. This now places emphasis on hurricane intensity and structure forecasts, which have barely improved and have little skill against climatological techniques. The unprecedented hurricane activity since 1995 in the North Atlantic, and particularly the last two seasons, has also generated intense debate in both scientific and political arenas on whether this is due to natural variability, or if it has an element of global climate change. Despite this growing importance to the Nation of improved hurricane intensity forecasts and projections, funding for theoretical and applied hurricane research remains completely inadequate. In light of the high level of impacts, the National Science Board established a Hurricane Research Task Force; the NOAA Science Advisory Board tasked a Hurricane Intensity Research Working Group (HIRWG), and the American Geophysical Union called a forum of experts to provide advice and recommendations. An ad hoc group of researchers also developed a prospectus for a hurricane process study. All have called for urgent increases in hurricane research and forecast technique improvement52. For example, the key recommendation of the HIRWG is: To reduce the error in 48-hour intensity forecasts for hurricane-strength storms by at least 10 kt (approximately one half of a Saffir-Simpson category) within the next five years, with an emphasis on improved forecasting of rapid intensification and decay, and decay and reintensification cycles. 5.a.2 Requirements Producing these essential improvements in forecasts and projections of hurricane intensity and structure is a complex undertaking that covers the entire theoretical and applied research spectrum and has three major components. First we need to improve critical deficiencies in our understanding of the manner in which the hurricane couples with the underlying ocean, of the interactions of the storm with its immediate environment, and of the small scale processes in the hurricane core that lead to rapid changes in hurricane intensity (both intensification and decay). These rapid changes have been identified by the Director of the National Hurricane Center as being the highest priority for improvement (HIRWRG report). Longer-term projections of hurricane impacts also require improvements in our understanding of how future hurricanes will respond to the atmospheric and oceanic changes associated with both natural variability and greenhouse warming. Second we need to improve the computer models that will both support the theoretical analysis and then carry them forward to become the mainstay of operational forecast and projection techniques.

52 Report of the Hurricane Intensity Research Working Group of the National Oceanic and Atmospheric Administration Science Advisory Board, July 2006. http://www.sab.noaa.gov/Working_Groups/Working_Groups.htm; Hurricanes and the U.S. Gulf Coast: Science and Sustainable Rebuilding, June 2006. American Geophysical Union. http://www.agu.org/report/hurricanes/

566

Recent research has shown the critical need for very high resolution to enable explicit modeling of the details of the high wind region, of the air-sea interaction and of the precipitation processes in the hurricane clouds over 3-5 day time periods. Projections of changes in hurricane intensity and structure also require climate projections with the capacity to resolve features of importance. But such high resolution comes at a very considerable cost in computer purchase and operation (potentially hundreds of millions of dollars). The model improvements will therefore need to include clever ways of producing the maximum improvement while minimizing the infrastructure costs. Third we need to substantially improve the observational base and the methods of bringing those observations into the forecast models. A deficiency of observations limits both our understanding of hurricane physics and our ability to verify and apply enhanced models. Here there are two major limitations – the cost of maintaining a targeted observing program, and our current incapacity to take detailed observations in critical areas, such as the high-wind layer near the ocean surface. Improvements here will require a careful combination of sophisticated techniques for assimilating data into the models, and targeted observations in the most critical regions. The targeted observations require a careful combination of current satellite, aircraft and surface observations with innovative new approaches. As with the models, this must include clever ways of minimizing the considerable cost whilst maximizing the improvements. This is obviously a major undertaking, which requires: • Collaborative planning and research by university, government and industry scientists; • Cooperative planning and support by federal and state agencies; • Utilization of the most advanced computing and observational resources; and, • Careful attention to ensuring that the research advances are transferred to operations in a manner that produces the maximum impact. 5.a.3 HiFi Program HiFi takes up the challenge proposed by the NSB and HIRWG reports for a focused program to enable critical details of the atmospheric convective processes, upper ocean heat, salt and momentum budgets, and the interactions between the upper ocean and the hurricane to be understood and included in next generation weather prediction models specifically tuned for hurricane prediction. This will occur through bringing together US and international expertise in hurricane and upper ocean physics, forecasters and climatologists in a carefully considered program of research, field studies and advanced numerical model and data assimilation development, Our major goal is: To make substantial and continuing improvements to the ocean and atmospheric models used to simulate hurricanes on both forecast and longer time scales. Achieving this goal will require focused research into understanding the fundamental processes together with improvements to observing programs, data assimilation, and modeling of key physical processes. Especially important are the modeling of clouds and the 2-way interaction of the hurricane with the underlying ocean. A fundamental set of issues concerns the nature of the air-sea interface and the coupled exchanges of heat, moisture and momentum between the atmosphere and ocean in hurricane cores. HiFi will build on and extend current research and operational programs that are currently dedicated to improved observing, understanding, forecasting and projection of hurricane intensity and structure. Some of the ongoing research activities include:

567

• Recent field programs such as RAINEX (NSF and NOAA) and CBLAST (ONR and NOAA); • The Weather Research and Forecast (WRF) model, which is coordinated by the National Center for Atmospheric Research (NCAR) as a cooperative activity between all government and private sector operational centers and academia, to enable rapid transfer of research improvements into operations; • Inclusion of a coupled ocean model with the WRF by NOAA and NCAR, which enables both studies and forecasting of the interaction between the ocean and hurricane circulation; • Major mobile observing facilities maintained by NOAA, NCAR and the USAF maintain with a long record of targeted hurricane observations; • Several new satellite observing facilities that have recently been launched, or are in the advanced development stage, for example the NASA A-Train satellites and the US-Taiwan COSMIC program; • Satellite altimeter missions supported by NASA and ONR that provide the basis for ocean heat content estimation and model initialization; • New generation observing systems, such as Unmanned Aerial Vehicles (UAVs), Autonomous Underwater Vehicles (AUVs) and Lagrangian Floats being tested by NASA, NOAA and ONR, which hold potential for continuous observations over long periods in hitherto unobservable regimes, such as the high wind and upper ocean layer;. • The WRF model has been recently extended to a Nested Regional Climate Model mode, which brings a capacity for forecast-mode improvements to support improvements in hurricane activity projections, and vice-versa; • NASA and DOE have major computing facilities and NCAR is leading the establishment of a peta-scale computing facility that will increase our current capacity by up to 1000 times. These programs provide important capacity for observing and modeling the atmosphere and ocean within hurricanes, and the interactions between them. They provide the basis for establishment of HiFi, which will extend these facilities and promote developments towards a more skillful and efficient forecasting system that incorporates state-of-the-art research advances with affordable observing and forecasting facilities. HiFi will extend this existing base to provide: • Improvements in understanding and models for, research, forecasting and impact projections; • Testing and development of new observing platforms and instruments, together with targeting and data assimilation techniques to maximize the utilization of observing systems; • Transition of research results into operational use. This is a major effort that will require coordination of researchers dedicated to improved understanding, with model and observing system developers, and with careful attention to the requirements of forecasters and vulnerable communities. Subject to the further planning efforts that are to be provided in the HiFi Science Strategy Reports, the work is to be broken down into three overlapping phases. Phase I will encompass an intensive research and development phase aimed at improving forecasts out to 5 days. This will include a series of intensive observation periods to build a sufficient database of ocean, interfacial and atmospheric fields to support the requirements of the theoretical research, analysis of efficient observing system approaches, and evaluation and validation of the next generation hurricane intensity models. The oil industry can uniquely contribute to this program through the use of their exploration and production platforms for critical atmospheric and oceanographic instrument deployment. Phase 2 will focus on hurricane impacts projections on longer time scales than 5 days, where ensemble forecasts of hurricane statistics are required. Regional climate modeling capability and capacity will be

568

improved to the level necessary to resolve and project the influences of natural variability and greenhouse warming on hurricane distributions and intensities. Coordinated theoretical and modeling studies will be used to provide the best scientific assessment of future trends in support of engineering design requirements, urban planning, and design of the relevant forecasting system. The timing and funding of this phase will have to be carefully considered at the initial science strategy and implementation planning meeting. Phase 3 will be the development and transition to operations of the next generation operational forecast system aimed at efficiently providing the information required for minimizing hurricane impacts on vulnerable coastal communities and facilities. The resulting improvements in intensity forecasts, and their translation into impacts prediction, will be of significant value to the coastal states, to the oil industry, to the reinsurance industry, and ultimately to the Nation. We estimate that this phase will commence during the first 5 years, but will extend for a further five years to ensure complete use of relevant research results. 5.a.4 Process

5.a.4.1 Program Leadership The magnitude of this integrated research and development effort requires leadership by a major national institution. We recommend that NCAR take on this leadership role, with the related calls for research and development proposals and distribution of awarded funds being managed by the University Corporation for Atmospheric Research (UCAR). NCAR already coordinates the development, support and operational transfer of the WRF model, which will be the mainstay of the HiFi program. NCAR also is a world leader in climate modeling and projection and has recently combined the WRF and Community Climate models to provide a capacity for resolving hurricane and other severe weather responses to global change. The capacity to undertake leading-edge experiments with these models will be given a major boost by NCAR’s moves towards a peta-scale computing facility. A national field operations facility also is maintained by NCAR’s Earth Observing Laboratory, which includes radars, surface instrumentation, and aircraft facilities. While the work will be lead and coordinated by UCAR/NCAR, the bulk of the research will be done by university collaborators and close coordination would be maintained with relevant government laboratories and offices. Selection of Dr. G. Holland and Professor R. Lukas as Principle Investigators will facilitate this coordination. Dr. Holland will lead the coordination of the UCAR/NCAR activities and serve as the lead atmospheric scientist. Professor Lukas will coordinate the academic program and serve as the lead oceanographic scientist. Dr Holland and Professor Lukas co-Chair a Steering Committee comprised of representatives of the major research and forecast organizations involved in HiFi. This steering committee will meet at regular periods to monitor progress with HiFi and advise on major directions and outcomes. Because of the high-level of impact on Gulf and Atlantic communities, Dr. Alexander Soloviev of Nova Southeastern University Oceanographic Center (NSU OC) in Dania Beach, Florida will serve on the HiFi Steering Committee to help coordinate with local communities and particularly the Coastal Ocean Observing System community. An Interim Project Office will be established at NSU OC in order to coordinate initial stages of the project development. We also propose that Dr. Frank Marks of the Hurricane Research Division in the Miami NOAA Atlantic Oceanic and Meteorological Laboratory take the lead in advising the committee on operational hurricane observing system developments.

569

5.a.4.2 Implementation This is an ambitious and complex program, for which this prospectus can provide only a minimum of detail. We propose a staged process for establishment and operation of the program and justification of the funding. Stage 1 will bring together the major potential scientific contributors, forecasters and community representatives to coordinate a substantial proposal, which will provide program details, expected major collaborators, coordination process and major milestones for the first five years. This will include development of a draft collaboration agreement between the Department of the Interior and UCAR/NCAR and could be completed in 3-6 months. A grant of $200,000 will be required to cover the related expenses, including salaries of the PIs and travel costs for focused meetings and the development of the HiFi Science Strategy Report and related documents. Stage 2 will be review of the proposed research program coordinated by a leading national body. While there are a number of groups that could undertake this, we recommend the National Research Council (jointly though its Board on Atmospheric Sciences and Climate and its Ocean Studies Board). This group would be tasked to assign a committee of experts to review the proposed work as provided in the HiFi Science Strategy Report and related documents, provide recommendations on the scope and activities and recommend continuation or cancellation to the Department of the Interior. Stage 3 will follow approval of the full program and will consist of the execution of the first five year research and development plan. We recommend that the NRC expert committee be tasked with an annual review of progress with this plan. Stage 4 will be development of a comprehensive operational plan in the 4th year of the program. This will detail a further 5-year plan for the full transfer of research and development results to a new generation operational observing and forecast system. It will include recommendations for carry-on applied research. We recommend that this be submitted to a committee chaired by the Director of the National Hurricane Center and consisting of representatives from NOAA, coastal community organizations, the offshore oil industry and academia. Stage 5 will be the execution of the second 5-year program to establish the next generation forecast and observing system.